Unverified Commit 5df80590 authored by Nicolas Patry's avatar Nicolas Patry Committed by GitHub
Browse files

Auto max prefill (#2797)

* Attempt at automatic max batch prefill.

* Taking into account number of shards.

* Adding more cards.

* Adding A100 + H100

* Adding a few more cards.

* Logprobs cost too much.

* h100 better name, and keep factor of 2

* Damn inflated sparse tflops.

* Typo in h100.

* Updated the flops calculation (checked with fvcore).

* chunking by default.

* Fix prefix caching for chat completion since we removed logprobs.

* More tests.

* Dropping all the prefill logprobs.

* Add a flag that enables users to get logprobs back.

* Repairing prompt token counting.

* Fixing a few tests.

* Remove some scaffolding.

* Attempting to reduces the issues (workarounds for now).
parent 8c3669b2
......@@ -3,68 +3,7 @@
"best_of_sequences": null,
"finish_reason": "eos_token",
"generated_tokens": 2,
"prefill": [
{
"id": 589,
"logprob": null,
"text": "def"
},
{
"id": 3226,
"logprob": -9.015625,
"text": " ge"
},
{
"id": 21017,
"logprob": -9.0859375,
"text": "ometric"
},
{
"id": 81,
"logprob": -0.25927734,
"text": "_"
},
{
"id": 6009,
"logprob": -2.2109375,
"text": "mean"
},
{
"id": 26,
"logprob": -0.2993164,
"text": "("
},
{
"id": 62,
"logprob": -5.671875,
"text": "L"
},
{
"id": 44,
"logprob": -3.0742188,
"text": ":"
},
{
"id": 1682,
"logprob": -0.6777344,
"text": " List"
},
{
"id": 77,
"logprob": -0.38354492,
"text": "["
},
{
"id": 1808,
"logprob": -0.91845703,
"text": "float"
},
{
"id": 10794,
"logprob": -2.5371094,
"text": "]):"
}
],
"prefill": [],
"seed": 0,
"tokens": [
{
......
......@@ -4,68 +4,7 @@
"best_of_sequences": null,
"finish_reason": "eos_token",
"generated_tokens": 2,
"prefill": [
{
"id": 589,
"logprob": null,
"text": "def"
},
{
"id": 3226,
"logprob": -8.9453125,
"text": " ge"
},
{
"id": 21017,
"logprob": -8.8515625,
"text": "ometric"
},
{
"id": 81,
"logprob": -0.22033691,
"text": "_"
},
{
"id": 6009,
"logprob": -1.2939453,
"text": "mean"
},
{
"id": 26,
"logprob": -0.25268555,
"text": "("
},
{
"id": 62,
"logprob": -4.796875,
"text": "L"
},
{
"id": 44,
"logprob": -3.796875,
"text": ":"
},
{
"id": 1682,
"logprob": -0.8066406,
"text": " List"
},
{
"id": 77,
"logprob": -0.22644043,
"text": "["
},
{
"id": 1808,
"logprob": -0.46166992,
"text": "float"
},
{
"id": 10794,
"logprob": -3.0253906,
"text": "]):"
}
],
"prefill": [],
"seed": null,
"tokens": [
{
......@@ -90,68 +29,7 @@
"best_of_sequences": null,
"finish_reason": "eos_token",
"generated_tokens": 2,
"prefill": [
{
"id": 589,
"logprob": null,
"text": "def"
},
{
"id": 3226,
"logprob": -8.9375,
"text": " ge"
},
{
"id": 21017,
"logprob": -8.8515625,
"text": "ometric"
},
{
"id": 81,
"logprob": -0.21826172,
"text": "_"
},
{
"id": 6009,
"logprob": -1.2871094,
"text": "mean"
},
{
"id": 26,
"logprob": -0.25390625,
"text": "("
},
{
"id": 62,
"logprob": -4.8085938,
"text": "L"
},
{
"id": 44,
"logprob": -3.7890625,
"text": ":"
},
{
"id": 1682,
"logprob": -0.8076172,
"text": " List"
},
{
"id": 77,
"logprob": -0.22302246,
"text": "["
},
{
"id": 1808,
"logprob": -0.46435547,
"text": "float"
},
{
"id": 10794,
"logprob": -3.0234375,
"text": "]):"
}
],
"prefill": [],
"seed": null,
"tokens": [
{
......@@ -176,68 +54,7 @@
"best_of_sequences": null,
"finish_reason": "eos_token",
"generated_tokens": 2,
"prefill": [
{
"id": 589,
"logprob": null,
"text": "def"
},
{
"id": 3226,
"logprob": -8.9453125,
"text": " ge"
},
{
"id": 21017,
"logprob": -8.8515625,
"text": "ometric"
},
{
"id": 81,
"logprob": -0.21813965,
"text": "_"
},
{
"id": 6009,
"logprob": -1.2744141,
"text": "mean"
},
{
"id": 26,
"logprob": -0.2512207,
"text": "("
},
{
"id": 62,
"logprob": -4.8046875,
"text": "L"
},
{
"id": 44,
"logprob": -3.7851562,
"text": ":"
},
{
"id": 1682,
"logprob": -0.81396484,
"text": " List"
},
{
"id": 77,
"logprob": -0.22570801,
"text": "["
},
{
"id": 1808,
"logprob": -0.46044922,
"text": "float"
},
{
"id": 10794,
"logprob": -3.0234375,
"text": "]):"
}
],
"prefill": [],
"seed": null,
"tokens": [
{
......@@ -262,68 +79,7 @@
"best_of_sequences": null,
"finish_reason": "eos_token",
"generated_tokens": 2,
"prefill": [
{
"id": 589,
"logprob": null,
"text": "def"
},
{
"id": 3226,
"logprob": -8.9453125,
"text": " ge"
},
{
"id": 21017,
"logprob": -8.8515625,
"text": "ometric"
},
{
"id": 81,
"logprob": -0.21960449,
"text": "_"
},
{
"id": 6009,
"logprob": -1.2890625,
"text": "mean"
},
{
"id": 26,
"logprob": -0.25073242,
"text": "("
},
{
"id": 62,
"logprob": -4.8085938,
"text": "L"
},
{
"id": 44,
"logprob": -3.8046875,
"text": ":"
},
{
"id": 1682,
"logprob": -0.8071289,
"text": " List"
},
{
"id": 77,
"logprob": -0.22570801,
"text": "["
},
{
"id": 1808,
"logprob": -0.46118164,
"text": "float"
},
{
"id": 10794,
"logprob": -3.0097656,
"text": "]):"
}
],
"prefill": [],
"seed": null,
"tokens": [
{
......
......@@ -3,88 +3,7 @@
"best_of_sequences": null,
"finish_reason": "eos_token",
"generated_tokens": 30,
"prefill": [
{
"id": 1,
"logprob": null,
"text": "<s>"
},
{
"id": 5235,
"logprob": -10.0625,
"text": "info"
},
{
"id": 29901,
"logprob": -3.2324219,
"text": ":"
},
{
"id": 13260,
"logprob": -10.625,
"text": "dav"
},
{
"id": 333,
"logprob": -0.08276367,
"text": "id"
},
{
"id": 8753,
"logprob": -7.5273438,
"text": "hol"
},
{
"id": 17559,
"logprob": -3.8476562,
"text": "tz"
},
{
"id": 763,
"logprob": -10.140625,
"text": "like"
},
{
"id": 10697,
"logprob": -10.1953125,
"text": "trees"
},
{
"id": 322,
"logprob": -2.5742188,
"text": "and"
},
{
"id": 756,
"logprob": -7.4882812,
"text": "has"
},
{
"id": 1023,
"logprob": -5.0507812,
"text": "two"
},
{
"id": 274,
"logprob": -5.3164062,
"text": "c"
},
{
"id": 1446,
"logprob": -0.6694336,
"text": "ats"
},
{
"id": 29889,
"logprob": -0.9995117,
"text": "."
},
{
"id": 29871,
"logprob": -4.2421875,
"text": ""
}
],
"prefill": [],
"seed": null,
"tokens": [
{
......
......@@ -3,23 +3,7 @@
"best_of_sequences": null,
"finish_reason": "length",
"generated_tokens": 10,
"prefill": [
{
"id": 1,
"logprob": null,
"text": "<s>"
},
{
"id": 3735,
"logprob": -8.5625,
"text": "Test"
},
{
"id": 2159,
"logprob": -10.78125,
"text": "request"
}
],
"prefill": [],
"seed": 0,
"tokens": [
{
......
......@@ -4,7030 +4,226 @@
"best_of_sequences": null,
"finish_reason": "length",
"generated_tokens": 10,
"prefill": [
{
"id": 1,
"logprob": null,
"text": "<s>"
},
{
"id": 1247,
"logprob": -5.2421875,
"text": "User"
},
{
"id": 28747,
"logprob": -6.9570312,
"text": ":"
},
{
"id": 32000,
"logprob": -16.234375,
"text": "<fake_token_around_image>"
},
{
"id": 32001,
"logprob": -18.96875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.1875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.46875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -17.578125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -16.8125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.296875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -20.21875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -16.421875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.84375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -23.265625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.421875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -15.28125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -17.734375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -20.34375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -21.3125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -21.03125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -20.421875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -16.03125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.046875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -22.765625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -23.609375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -20.421875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -20.4375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -20.875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -16.5,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.96875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.234375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -23.59375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -21.1875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -17.359375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -16.53125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.03125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -21.78125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.328125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -17.96875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -17.921875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.1875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.65625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -20.140625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.4375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -14.8828125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.453125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -23.203125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -17.421875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -23.671875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -23.0,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.75,
"text": "<image>"
},
{
"id": 32001,
"logprob": -17.078125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -17.640625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -17.046875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.40625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -20.578125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.328125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -17.125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.65625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.890625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -15.9921875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.15625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.84375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -20.15625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.96875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.703125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -17.921875,
"text": "<image>"
},
{
"id": 32000,
"logprob": -3.015625,
"text": "<fake_token_around_image>"
},
{
"id": 32001,
"logprob": -22.09375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.96875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -20.140625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -17.125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -17.796875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.3125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.65625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -16.3125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -20.046875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -20.828125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -15.8125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -16.25,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.953125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -21.59375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -20.515625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.171875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.296875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -16.703125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -20.46875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -21.125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -20.40625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -20.71875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -17.78125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -20.328125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -15.2734375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.84375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -17.875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -15.578125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.3125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -20.625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -20.140625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.34375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -17.796875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -17.046875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -16.5625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -20.859375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.140625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.546875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -14.1953125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.90625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.171875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -17.421875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -20.421875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -20.0,
"text": "<image>"
},
{
"id": 32001,
"logprob": -20.359375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.03125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -17.203125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -16.84375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.09375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -15.71875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.1875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.4375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.46875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -17.3125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -16.265625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.4375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -14.734375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -20.6875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -20.21875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.359375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.1875,
"text": "<image>"
},
{
"id": 32000,
"logprob": -3.2988281,
"text": "<fake_token_around_image>"
},
{
"id": 32001,
"logprob": -25.75,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.421875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.265625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -17.921875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -17.203125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -20.140625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -17.96875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -16.46875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.65625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.203125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -16.921875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -15.7421875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -16.375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -17.25,
"text": "<image>"
},
{
"id": 32001,
"logprob": -16.5625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.828125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.765625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -16.90625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.984375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.765625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.890625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -20.421875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.34375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -20.140625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.34375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.015625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -17.65625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -16.984375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -15.9453125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.53125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -20.078125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.953125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -15.4375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.8125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -15.4453125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -20.6875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -20.1875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.03125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -17.40625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -15.984375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.21875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.71875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.9375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -20.421875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.46875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.921875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -17.8125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.90625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -17.09375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -16.234375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -16.25,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.984375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.578125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -16.328125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -17.703125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -16.4375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.359375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -17.015625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -17.03125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -20.140625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.5625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.0,
"text": "<image>"
},
{
"id": 32000,
"logprob": -2.7207031,
"text": "<fake_token_around_image>"
},
{
"id": 32001,
"logprob": -23.34375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -22.203125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -21.015625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.578125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.1875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -20.296875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.5,
"text": "<image>"
},
{
"id": 32001,
"logprob": -16.46875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -20.890625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.390625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.96875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -16.0625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -17.265625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.53125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -20.5,
"text": "<image>"
},
{
"id": 32001,
"logprob": -21.5,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.671875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.765625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.671875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.203125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -21.03125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -21.34375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.3125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -20.609375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.6875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.921875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -20.9375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.46875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.0,
"text": "<image>"
},
{
"id": 32001,
"logprob": -17.40625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.640625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.59375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -20.25,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.828125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -15.796875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.765625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -15.6640625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -21.3125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.796875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.671875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.640625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -15.96875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -20.1875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.8125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -20.921875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -21.46875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -21.25,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.09375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -17.59375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -17.09375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -16.90625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -20.171875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -20.921875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -16.453125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.984375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -16.4375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -17.59375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -17.171875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -20.34375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.171875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.578125,
"text": "<image>"
},
{
"id": 32000,
"logprob": -3.0917969,
"text": "<fake_token_around_image>"
},
{
"id": 32001,
"logprob": -25.375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.921875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -20.6875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -17.921875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.75,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.71875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -21.453125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -16.734375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -20.875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -21.453125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -15.796875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -15.1328125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -17.125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.90625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -21.421875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -21.015625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -20.734375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -16.25,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.5,
"text": "<image>"
},
{
"id": 32001,
"logprob": -21.59375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -22.515625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -20.921875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.703125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -21.0,
"text": "<image>"
},
{
"id": 32001,
"logprob": -16.984375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -17.53125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -17.9375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -22.1875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.75,
"text": "<image>"
},
{
"id": 32001,
"logprob": -16.390625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.421875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -20.265625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -22.296875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.484375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -15.3984375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.75,
"text": "<image>"
},
{
"id": 32001,
"logprob": -14.6484375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -21.609375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.828125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -20.84375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -17.015625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -16.40625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -21.046875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -21.234375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -17.125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -21.5,
"text": "<image>"
},
{
"id": 32001,
"logprob": -20.0,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.78125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -16.375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -16.875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -16.703125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -13.6171875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -15.375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -17.5,
"text": "<image>"
},
{
"id": 32001,
"logprob": -21.921875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -15.640625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -16.46875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -16.421875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.890625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -17.890625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -17.40625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -20.390625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.1875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -15.9609375,
"text": "<image>"
},
{
"id": 32000,
"logprob": -2.0332031,
"text": "<fake_token_around_image>"
},
{
"id": 12018,
"logprob": -12.078125,
"text": "Write"
},
{
"id": 528,
"logprob": -10.09375,
"text": "me"
},
{
"id": 264,
"logprob": -0.103393555,
"text": "a"
},
{
"id": 2485,
"logprob": -4.5742188,
"text": "short"
},
{
"id": 2838,
"logprob": -0.23815918,
"text": "story"
},
{
"id": 32002,
"logprob": -10.9765625,
"text": "<end_of_utterance>"
},
{
"id": 259,
"logprob": -20.34375,
"text": " "
},
{
"id": 13,
"logprob": -8.53125,
"text": "\n"
},
{
"id": 7226,
"logprob": -10.4765625,
"text": "Ass"
},
{
"id": 11143,
"logprob": -13.6015625,
"text": "istant"
},
{
"id": 28747,
"logprob": -0.008514404,
"text": ":"
}
],
"seed": null,
"tokens": [
{
"id": 330,
"logprob": -0.09289551,
"special": false,
"text": " A"
},
{
"id": 13088,
"logprob": -0.6743164,
"special": false,
"text": " chicken"
},
{
"id": 349,
"logprob": -0.31396484,
"special": false,
"text": " is"
},
{
"id": 6398,
"logprob": -0.051727295,
"special": false,
"text": " sitting"
},
{
"id": 356,
"logprob": -0.34448242,
"special": false,
"text": " on"
},
{
"id": 264,
"logprob": -0.1194458,
"special": false,
"text": " a"
},
{
"id": 17972,
"logprob": -0.03237915,
"special": false,
"text": " pile"
},
{
"id": 302,
"logprob": -0.00018751621,
"special": false,
"text": " of"
},
{
"id": 2445,
"logprob": -0.07043457,
"special": false,
"text": " money"
},
{
"id": 28723,
"logprob": -0.00422287,
"special": false,
"text": "."
}
],
"top_tokens": null
},
"generated_text": " A chicken is sitting on a pile of money."
},
{
"details": {
"best_of_sequences": null,
"finish_reason": "length",
"generated_tokens": 10,
"prefill": [
{
"id": 1,
"logprob": null,
"text": "<s>"
},
{
"id": 1247,
"logprob": -5.2382812,
"text": "User"
},
{
"id": 28747,
"logprob": -6.9492188,
"text": ":"
},
{
"id": 32000,
"logprob": -16.234375,
"text": "<fake_token_around_image>"
},
{
"id": 32001,
"logprob": -18.984375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.171875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.46875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -17.578125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -16.8125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.296875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -20.21875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -16.421875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.84375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -23.25,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.421875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -15.28125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -17.734375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -20.34375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -21.296875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -21.03125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -20.421875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -16.03125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.046875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -22.765625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -23.609375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -20.421875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -20.421875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -20.875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -16.5,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.96875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.234375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -23.59375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -21.1875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -17.359375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -16.53125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.03125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -21.78125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.328125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -17.96875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.859375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -17.921875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.1875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.65625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -20.140625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.4375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -14.8828125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.453125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -23.21875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -17.421875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -23.671875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -23.0,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.75,
"text": "<image>"
},
{
"id": 32001,
"logprob": -17.078125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -17.640625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -17.046875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.421875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -20.578125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.328125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -17.125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.65625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -16.0,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.15625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.84375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -20.15625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.96875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.703125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -17.921875,
"text": "<image>"
},
{
"id": 32000,
"logprob": -3.015625,
"text": "<fake_token_around_image>"
},
{
"id": 32001,
"logprob": -22.109375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.96875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -20.125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -17.125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -17.8125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.3125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.65625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -16.3125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -20.046875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -20.828125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -15.8046875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -16.25,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.953125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -21.59375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -20.515625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.171875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.296875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -16.71875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -20.46875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -21.125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -20.40625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -20.734375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -17.765625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -20.328125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -15.2734375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.84375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -17.875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -15.578125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.3125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -20.625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -20.140625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.34375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -17.8125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -17.046875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -16.578125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -20.859375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.15625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.546875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -14.1953125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.90625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.1875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -17.421875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -20.421875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -20.0,
"text": "<image>"
},
{
"id": 32001,
"logprob": -20.359375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.03125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -17.203125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -16.84375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.09375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -15.71875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.203125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.4375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.46875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -17.3125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -16.265625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.453125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -14.734375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -20.6875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -20.21875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.359375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.1875,
"text": "<image>"
},
{
"id": 32000,
"logprob": -3.2988281,
"text": "<fake_token_around_image>"
},
{
"id": 32001,
"logprob": -25.75,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.421875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.265625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -17.90625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -17.203125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -20.140625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -17.96875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -16.453125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.65625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.203125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -16.921875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -15.75,
"text": "<image>"
},
{
"id": 32001,
"logprob": -16.375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -17.25,
"text": "<image>"
},
{
"id": 32001,
"logprob": -16.5625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.828125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.765625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -16.90625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.984375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.765625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.890625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -20.421875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.34375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -20.140625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.34375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.015625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -17.671875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -16.984375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -15.953125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.53125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -20.078125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.953125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -15.4453125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.828125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -15.453125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -20.6875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -20.1875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.03125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -17.40625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -15.984375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.21875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.71875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.9375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -20.421875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.46875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.90625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -17.8125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.90625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -17.09375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -16.234375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -16.25,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.984375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.59375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -16.328125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -17.703125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -16.4375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.359375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -17.015625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -17.03125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -20.140625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.5625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.0,
"text": "<image>"
},
{
"id": 32000,
"logprob": -2.7207031,
"text": "<fake_token_around_image>"
},
{
"id": 32001,
"logprob": -23.34375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -22.203125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -21.015625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.578125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.1875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -20.296875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.484375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -16.46875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -20.890625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.390625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.96875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -16.078125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -17.28125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.53125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -20.5,
"text": "<image>"
},
{
"id": 32001,
"logprob": -21.5,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.6875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.765625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.671875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.203125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -21.03125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -21.328125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.3125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -20.609375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.6875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.921875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -20.9375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.46875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.0,
"text": "<image>"
},
{
"id": 32001,
"logprob": -17.40625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.640625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.59375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -20.25,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.828125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -15.796875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.765625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -15.6640625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -21.3125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.8125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.671875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.640625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -15.96875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -20.1875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.8125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -20.921875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -21.46875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -21.25,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.09375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -17.59375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -17.09375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -16.90625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -20.171875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -20.921875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -16.453125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.984375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -16.4375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -17.59375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -17.171875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -20.34375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.171875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.578125,
"text": "<image>"
},
{
"id": 32000,
"logprob": -3.0917969,
"text": "<fake_token_around_image>"
},
{
"id": 32001,
"logprob": -25.375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.921875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -20.6875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -17.921875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.734375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.71875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -21.453125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -16.734375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -20.875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -21.453125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -15.796875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -15.1328125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -17.125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.90625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -21.421875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -21.015625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -20.734375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -16.25,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.5,
"text": "<image>"
},
{
"id": 32001,
"logprob": -21.59375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -22.515625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -20.921875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.703125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -21.0,
"text": "<image>"
},
{
"id": 32001,
"logprob": -16.984375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -17.53125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -17.921875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -22.1875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.75,
"text": "<image>"
},
{
"id": 32001,
"logprob": -16.375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.4375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -20.265625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -22.296875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.484375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -15.390625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.75,
"text": "<image>"
},
{
"id": 32001,
"logprob": -14.6484375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -21.609375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.828125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -20.828125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -17.015625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -16.40625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -21.046875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -21.234375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -17.140625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -21.515625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -20.0,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.78125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -16.375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -16.890625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -16.703125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -13.625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -15.375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -17.515625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -21.921875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -15.640625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -16.46875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -16.421875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.890625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -17.890625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -17.40625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -20.390625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.1875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -15.9609375,
"text": "<image>"
},
{
"id": 32000,
"logprob": -2.0332031,
"text": "<fake_token_around_image>"
},
{
"id": 12018,
"logprob": -12.078125,
"text": "Write"
},
{
"id": 528,
"logprob": -10.109375,
"text": "me"
},
{
"id": 264,
"logprob": -0.103515625,
"text": "a"
},
{
"id": 2485,
"logprob": -4.5664062,
"text": "short"
},
{
"id": 2838,
"logprob": -0.23864746,
"text": "story"
},
{
"id": 32002,
"logprob": -10.9609375,
"text": "<end_of_utterance>"
},
{
"id": 259,
"logprob": -20.34375,
"text": " "
},
{
"id": 13,
"logprob": -8.5546875,
"text": "\n"
},
{
"id": 7226,
"logprob": -10.484375,
"text": "Ass"
},
{
"id": 11143,
"logprob": -13.6015625,
"text": "istant"
},
{
"id": 28747,
"logprob": -0.008308411,
"text": ":"
}
],
"seed": null,
"tokens": [
{
"id": 330,
"logprob": -0.09448242,
"special": false,
"text": " A"
},
{
"id": 13088,
"logprob": -0.6743164,
"special": false,
"text": " chicken"
},
{
"id": 349,
"logprob": -0.31201172,
"special": false,
"text": " is"
},
{
"id": 6398,
"logprob": -0.051635742,
"special": false,
"text": " sitting"
},
{
"id": 356,
"logprob": -0.34033203,
"special": false,
"text": " on"
},
{
"id": 264,
"logprob": -0.1194458,
"special": false,
"text": " a"
},
{
"id": 17972,
"logprob": -0.032562256,
"special": false,
"text": " pile"
},
{
"id": 302,
"logprob": -0.00018763542,
"special": false,
"text": " of"
},
{
"id": 2445,
"logprob": -0.07122803,
"special": false,
"text": " money"
},
{
"id": 28723,
"logprob": -0.0041007996,
"special": false,
"text": "."
}
],
"top_tokens": null
},
"generated_text": " A chicken is sitting on a pile of money."
},
{
"details": {
"best_of_sequences": null,
"finish_reason": "length",
"generated_tokens": 10,
"prefill": [
{
"id": 1,
"logprob": null,
"text": "<s>"
},
{
"id": 1247,
"logprob": -5.2382812,
"text": "User"
},
{
"id": 28747,
"logprob": -6.9492188,
"text": ":"
},
{
"id": 32000,
"logprob": -16.234375,
"text": "<fake_token_around_image>"
},
{
"id": 32001,
"logprob": -18.984375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.171875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.46875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -17.578125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -16.8125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.296875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -20.21875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -16.421875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.84375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -23.25,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.421875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -15.28125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -17.734375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -20.34375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -21.296875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -21.03125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -20.421875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -16.03125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.046875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -22.765625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -23.609375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -20.421875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -20.421875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -20.875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -16.5,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.96875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.234375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -23.59375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -21.1875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -17.359375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -16.53125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.03125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -21.78125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.328125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -17.96875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.859375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -17.921875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.1875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.65625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -20.140625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.4375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -14.8828125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.453125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -23.21875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -17.421875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -23.671875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -23.0,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.75,
"text": "<image>"
},
{
"id": 32001,
"logprob": -17.078125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -17.640625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -17.046875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.421875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -20.578125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.328125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -17.125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.65625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -16.0,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.15625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.84375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -20.15625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.96875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.703125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -17.921875,
"text": "<image>"
},
{
"id": 32000,
"logprob": -3.015625,
"text": "<fake_token_around_image>"
},
{
"id": 32001,
"logprob": -22.109375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.96875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -20.125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -17.125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -17.8125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.3125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.65625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -16.3125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -20.046875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -20.828125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -15.8046875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -16.25,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.953125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -21.59375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -20.515625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.171875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.296875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -16.71875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -20.46875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -21.125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -20.40625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -20.734375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -17.765625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -20.328125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -15.2734375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.84375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -17.875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -15.578125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.3125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -20.625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -20.140625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.34375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -17.8125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -17.046875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -16.578125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -20.859375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.15625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.546875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -14.1953125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.90625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.1875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -17.421875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -20.421875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -20.0,
"text": "<image>"
},
{
"id": 32001,
"logprob": -20.359375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.03125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -17.203125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -16.84375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.09375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -15.71875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.203125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.4375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.46875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -17.3125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -16.265625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.453125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -14.734375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -20.6875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -20.21875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.359375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.1875,
"text": "<image>"
},
{
"id": 32000,
"logprob": -3.2988281,
"text": "<fake_token_around_image>"
},
{
"id": 32001,
"logprob": -25.75,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.421875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.265625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -17.90625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -17.203125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -20.140625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -17.96875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -16.453125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.65625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.203125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -16.921875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -15.75,
"text": "<image>"
},
{
"id": 32001,
"logprob": -16.375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -17.25,
"text": "<image>"
},
{
"id": 32001,
"logprob": -16.5625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.828125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.765625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -16.90625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.984375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.765625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.890625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -20.421875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.34375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -20.140625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.34375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.015625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -17.671875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -16.984375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -15.953125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.53125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -20.078125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.953125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -15.4453125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.828125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -15.453125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -20.6875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -20.1875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.03125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -17.40625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -15.984375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.21875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.71875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.9375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -20.421875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.46875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.90625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -17.8125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.90625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -17.09375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -16.234375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -16.25,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.984375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.59375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -16.328125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -17.703125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -16.4375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.359375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -17.015625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -17.03125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -20.140625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.5625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.0,
"text": "<image>"
},
{
"id": 32000,
"logprob": -2.7207031,
"text": "<fake_token_around_image>"
},
{
"id": 32001,
"logprob": -23.34375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -22.203125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -21.015625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.578125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.1875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -20.296875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.484375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -16.46875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -20.890625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.390625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.96875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -16.078125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -17.28125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.53125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -20.5,
"text": "<image>"
},
{
"id": 32001,
"logprob": -21.5,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.6875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.765625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.671875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.203125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -21.03125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -21.328125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.3125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -20.609375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.6875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.921875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -20.9375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.46875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.0,
"text": "<image>"
},
{
"id": 32001,
"logprob": -17.40625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.640625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.59375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -20.25,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.828125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -15.796875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.765625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -15.6640625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -21.3125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.8125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.671875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.640625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -15.96875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -20.1875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.8125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -20.921875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -21.46875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -21.25,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.09375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -17.59375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -17.09375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -16.90625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -20.171875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -20.921875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -16.453125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.984375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -16.4375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -17.59375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -17.171875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -20.34375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.171875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.578125,
"text": "<image>"
},
{
"id": 32000,
"logprob": -3.0917969,
"text": "<fake_token_around_image>"
},
{
"id": 32001,
"logprob": -25.375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.921875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -20.6875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -17.921875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.734375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.71875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -21.453125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -16.734375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -20.875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -21.453125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -15.796875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -15.1328125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -17.125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.90625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -21.421875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -21.015625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -20.734375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -16.25,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.5,
"text": "<image>"
},
{
"id": 32001,
"logprob": -21.59375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -22.515625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -20.921875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.703125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -21.0,
"text": "<image>"
},
{
"id": 32001,
"logprob": -16.984375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -17.53125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -17.921875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -22.1875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.75,
"text": "<image>"
},
{
"id": 32001,
"logprob": -16.375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.4375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -20.265625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -22.296875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.484375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -15.390625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.75,
"text": "<image>"
},
{
"id": 32001,
"logprob": -14.6484375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -21.609375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.828125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -20.828125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -17.015625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -16.40625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -21.046875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -21.234375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -17.140625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -21.515625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -20.0,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.78125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -16.375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -16.890625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -16.703125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -13.625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -15.375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -17.515625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -21.921875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -15.640625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -16.46875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -16.421875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.890625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -17.890625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -17.40625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -20.390625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.1875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -15.9609375,
"text": "<image>"
},
{
"id": 32000,
"logprob": -2.0332031,
"text": "<fake_token_around_image>"
},
{
"id": 12018,
"logprob": -12.078125,
"text": "Write"
},
{
"id": 528,
"logprob": -10.109375,
"text": "me"
},
{
"id": 264,
"logprob": -0.103515625,
"text": "a"
},
{
"id": 2485,
"logprob": -4.5664062,
"text": "short"
},
{
"id": 2838,
"logprob": -0.23864746,
"text": "story"
},
{
"id": 32002,
"logprob": -10.9609375,
"text": "<end_of_utterance>"
},
{
"id": 259,
"logprob": -20.34375,
"text": " "
},
{
"id": 13,
"logprob": -8.5546875,
"text": "\n"
},
{
"id": 7226,
"logprob": -10.484375,
"text": "Ass"
},
{
"id": 11143,
"logprob": -13.6015625,
"text": "istant"
},
{
"id": 28747,
"logprob": -0.008308411,
"text": ":"
}
],
"seed": null,
"tokens": [
{
"id": 330,
"logprob": -0.09448242,
"special": false,
"text": " A"
},
{
"id": 13088,
"logprob": -0.6743164,
"special": false,
"text": " chicken"
},
{
"id": 349,
"logprob": -0.31201172,
"special": false,
"text": " is"
},
{
"id": 6398,
"logprob": -0.051635742,
"special": false,
"text": " sitting"
},
{
"id": 356,
"logprob": -0.34033203,
"special": false,
"text": " on"
},
{
"id": 264,
"logprob": -0.1194458,
"special": false,
"text": " a"
},
{
"id": 17972,
"logprob": -0.032562256,
"special": false,
"text": " pile"
},
{
"id": 302,
"logprob": -0.00018787384,
"special": false,
"text": " of"
},
{
"id": 2445,
"logprob": -0.07122803,
"special": false,
"text": " money"
},
{
"id": 28723,
"logprob": -0.0041007996,
"special": false,
"text": "."
}
],
"top_tokens": null
},
"generated_text": " A chicken is sitting on a pile of money."
},
{
"details": {
"best_of_sequences": null,
"finish_reason": "length",
"generated_tokens": 10,
"prefill": [
{
"id": 1,
"logprob": null,
"text": "<s>"
},
{
"id": 1247,
"logprob": -5.2382812,
"text": "User"
},
{
"id": 28747,
"logprob": -6.9492188,
"text": ":"
},
{
"id": 32000,
"logprob": -16.234375,
"text": "<fake_token_around_image>"
},
{
"id": 32001,
"logprob": -18.984375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.171875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.46875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -17.578125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -16.8125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.296875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -20.21875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -16.421875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.84375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -23.25,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.421875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -15.28125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -17.734375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -20.34375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -21.296875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -21.03125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -20.421875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -16.03125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.046875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -22.765625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -23.609375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -20.421875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -20.421875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -20.875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -16.5,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.96875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.234375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -23.59375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -21.1875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -17.359375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -16.53125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.03125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -21.78125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.328125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -17.96875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.859375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -17.921875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.1875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.65625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -20.140625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.4375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -14.8828125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.453125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -23.21875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -17.421875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -23.671875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -23.0,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.75,
"text": "<image>"
},
{
"id": 32001,
"logprob": -17.078125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -17.640625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -17.046875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.421875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -20.578125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.328125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -17.125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.65625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -16.0,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.15625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.84375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -20.15625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.96875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.703125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -17.921875,
"text": "<image>"
},
{
"id": 32000,
"logprob": -3.015625,
"text": "<fake_token_around_image>"
},
{
"id": 32001,
"logprob": -22.109375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.96875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -20.125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -17.125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -17.8125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.3125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.65625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -16.3125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -20.046875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -20.828125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -15.8046875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -16.25,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.953125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -21.59375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -20.515625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.171875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.296875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -16.71875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -20.46875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -21.125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -20.40625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -20.734375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -17.765625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -20.328125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -15.2734375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.84375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -17.875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -15.578125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.3125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -20.625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -20.140625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.34375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -17.8125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -17.046875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -16.578125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -20.859375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.15625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.546875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -14.1953125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.90625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.1875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -17.421875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -20.421875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -20.0,
"text": "<image>"
},
{
"id": 32001,
"logprob": -20.359375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.03125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -17.203125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -16.84375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.09375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -15.71875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.203125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.4375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.46875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -17.3125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -16.265625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.453125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -14.734375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -20.6875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -20.21875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.359375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.1875,
"text": "<image>"
},
{
"id": 32000,
"logprob": -3.2988281,
"text": "<fake_token_around_image>"
},
{
"id": 32001,
"logprob": -25.75,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.421875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.265625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -17.90625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -17.203125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -20.140625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -17.96875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -16.453125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.65625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.203125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -16.921875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -15.75,
"text": "<image>"
},
{
"id": 32001,
"logprob": -16.375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -17.25,
"text": "<image>"
},
{
"id": 32001,
"logprob": -16.5625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.828125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.765625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -16.90625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.984375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.765625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.890625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -20.421875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.34375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -20.140625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.34375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.015625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -17.671875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -16.984375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -15.953125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.53125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -20.078125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.953125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -15.4453125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.828125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -15.453125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -20.6875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -20.1875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.03125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -17.40625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -15.984375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.21875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.71875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.9375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -20.421875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.46875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.90625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -17.8125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.90625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -17.09375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -16.234375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -16.25,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.984375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.59375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -16.328125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -17.703125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -16.4375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.359375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -17.015625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -17.03125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -20.140625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.5625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.0,
"text": "<image>"
},
{
"id": 32000,
"logprob": -2.7207031,
"text": "<fake_token_around_image>"
},
{
"id": 32001,
"logprob": -23.34375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -22.203125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -21.015625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.578125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.1875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -20.296875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.484375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -16.46875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -20.890625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.390625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.96875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -16.078125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -17.28125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.53125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -20.5,
"text": "<image>"
},
{
"id": 32001,
"logprob": -21.5,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.6875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.765625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.671875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.203125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -21.03125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -21.328125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.3125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -20.609375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.6875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.921875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -20.9375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.46875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.0,
"text": "<image>"
},
{
"id": 32001,
"logprob": -17.40625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.640625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.59375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -20.25,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.828125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -15.796875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.765625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -15.6640625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -21.3125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.8125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.671875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.640625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -15.96875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -20.1875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.8125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -20.921875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -21.46875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -21.25,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.09375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -17.59375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -17.09375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -16.90625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -20.171875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -20.921875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -16.453125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.984375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -16.4375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -17.59375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -17.171875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -20.34375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.171875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.578125,
"text": "<image>"
},
{
"id": 32000,
"logprob": -3.0917969,
"text": "<fake_token_around_image>"
},
{
"id": 32001,
"logprob": -25.375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.921875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -20.6875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -17.921875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.734375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.71875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -21.453125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -16.734375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -20.875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -21.453125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -15.796875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -15.1328125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -17.125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.90625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -21.421875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -21.015625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -20.734375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -16.25,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.5,
"text": "<image>"
},
{
"id": 32001,
"logprob": -21.59375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -22.515625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -20.921875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.703125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -21.0,
"text": "<image>"
},
{
"id": 32001,
"logprob": -16.984375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -17.53125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -17.921875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -22.1875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.75,
"text": "<image>"
},
{
"id": 32001,
"logprob": -16.375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.4375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -20.265625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -22.296875,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.484375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -15.390625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -19.75,
"text": "<image>"
},
{
"id": 32001,
"logprob": -14.6484375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -21.609375,
"text": "<image>"
},
{
"id": 32001,
"logprob": -18.828125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -20.828125,
"text": "<image>"
},
{
"id": 32001,
"logprob": -17.015625,
"text": "<image>"
},
{
"id": 32001,
"logprob": -16.40625,
"text": "<image>"
},
"prefill": [],
"seed": null,
"tokens": [
{
"id": 32001,
"logprob": -21.046875,
"text": "<image>"
"id": 330,
"logprob": -0.09289551,
"special": false,
"text": " A"
},
{
"id": 32001,
"logprob": -21.234375,
"text": "<image>"
"id": 13088,
"logprob": -0.6743164,
"special": false,
"text": " chicken"
},
{
"id": 32001,
"logprob": -17.140625,
"text": "<image>"
"id": 349,
"logprob": -0.31396484,
"special": false,
"text": " is"
},
{
"id": 32001,
"logprob": -21.515625,
"text": "<image>"
"id": 6398,
"logprob": -0.051727295,
"special": false,
"text": " sitting"
},
{
"id": 32001,
"logprob": -20.0,
"text": "<image>"
"id": 356,
"logprob": -0.34448242,
"special": false,
"text": " on"
},
{
"id": 32001,
"logprob": -18.78125,
"text": "<image>"
"id": 264,
"logprob": -0.1194458,
"special": false,
"text": " a"
},
{
"id": 32001,
"logprob": -16.375,
"text": "<image>"
"id": 17972,
"logprob": -0.03237915,
"special": false,
"text": " pile"
},
{
"id": 32001,
"logprob": -16.890625,
"text": "<image>"
"id": 302,
"logprob": -0.00018751621,
"special": false,
"text": " of"
},
{
"id": 32001,
"logprob": -16.703125,
"text": "<image>"
"id": 2445,
"logprob": -0.07043457,
"special": false,
"text": " money"
},
{
"id": 32001,
"logprob": -13.625,
"text": "<image>"
},
"id": 28723,
"logprob": -0.00422287,
"special": false,
"text": "."
}
],
"top_tokens": null
},
"generated_text": " A chicken is sitting on a pile of money."
},
{
"details": {
"best_of_sequences": null,
"finish_reason": "length",
"generated_tokens": 10,
"prefill": [],
"seed": null,
"tokens": [
{
"id": 32001,
"logprob": -15.375,
"text": "<image>"
"id": 330,
"logprob": -0.09448242,
"special": false,
"text": " A"
},
{
"id": 32001,
"logprob": -17.515625,
"text": "<image>"
"id": 13088,
"logprob": -0.6743164,
"special": false,
"text": " chicken"
},
{
"id": 32001,
"logprob": -21.921875,
"text": "<image>"
"id": 349,
"logprob": -0.31201172,
"special": false,
"text": " is"
},
{
"id": 32001,
"logprob": -15.640625,
"text": "<image>"
"id": 6398,
"logprob": -0.051635742,
"special": false,
"text": " sitting"
},
{
"id": 32001,
"logprob": -16.46875,
"text": "<image>"
"id": 356,
"logprob": -0.34033203,
"special": false,
"text": " on"
},
{
"id": 32001,
"logprob": -16.421875,
"text": "<image>"
"id": 264,
"logprob": -0.1194458,
"special": false,
"text": " a"
},
{
"id": 32001,
"logprob": -19.890625,
"text": "<image>"
"id": 17972,
"logprob": -0.032562256,
"special": false,
"text": " pile"
},
{
"id": 32001,
"logprob": -17.890625,
"text": "<image>"
"id": 302,
"logprob": -0.00018763542,
"special": false,
"text": " of"
},
{
"id": 32001,
"logprob": -17.40625,
"text": "<image>"
"id": 2445,
"logprob": -0.07122803,
"special": false,
"text": " money"
},
{
"id": 32001,
"logprob": -20.390625,
"text": "<image>"
},
"id": 28723,
"logprob": -0.0041007996,
"special": false,
"text": "."
}
],
"top_tokens": null
},
"generated_text": " A chicken is sitting on a pile of money."
},
{
"details": {
"best_of_sequences": null,
"finish_reason": "length",
"generated_tokens": 10,
"prefill": [],
"seed": null,
"tokens": [
{
"id": 32001,
"logprob": -19.1875,
"text": "<image>"
"id": 330,
"logprob": -0.09448242,
"special": false,
"text": " A"
},
{
"id": 32001,
"logprob": -15.9609375,
"text": "<image>"
"id": 13088,
"logprob": -0.6743164,
"special": false,
"text": " chicken"
},
{
"id": 32000,
"logprob": -2.0332031,
"text": "<fake_token_around_image>"
"id": 349,
"logprob": -0.31201172,
"special": false,
"text": " is"
},
{
"id": 12018,
"logprob": -12.078125,
"text": "Write"
"id": 6398,
"logprob": -0.051635742,
"special": false,
"text": " sitting"
},
{
"id": 528,
"logprob": -10.109375,
"text": "me"
"id": 356,
"logprob": -0.34033203,
"special": false,
"text": " on"
},
{
"id": 264,
"logprob": -0.103515625,
"text": "a"
},
{
"id": 2485,
"logprob": -4.5664062,
"text": "short"
},
{
"id": 2838,
"logprob": -0.23864746,
"text": "story"
},
{
"id": 32002,
"logprob": -10.9609375,
"text": "<end_of_utterance>"
},
{
"id": 259,
"logprob": -20.34375,
"text": " "
"logprob": -0.1194458,
"special": false,
"text": " a"
},
{
"id": 13,
"logprob": -8.5546875,
"text": "\n"
"id": 17972,
"logprob": -0.032562256,
"special": false,
"text": " pile"
},
{
"id": 7226,
"logprob": -10.484375,
"text": "Ass"
"id": 302,
"logprob": -0.00018787384,
"special": false,
"text": " of"
},
{
"id": 11143,
"logprob": -13.6015625,
"text": "istant"
"id": 2445,
"logprob": -0.07122803,
"special": false,
"text": " money"
},
{
"id": 28747,
"logprob": -0.008308411,
"text": ":"
"id": 28723,
"logprob": -0.0041007996,
"special": false,
"text": "."
}
],
"top_tokens": null
},
"generated_text": " A chicken is sitting on a pile of money."
},
{
"details": {
"best_of_sequences": null,
"finish_reason": "length",
"generated_tokens": 10,
"prefill": [],
"seed": null,
"tokens": [
{
......
......@@ -3,23 +3,7 @@
"best_of_sequences": null,
"finish_reason": "stop_sequence",
"generated_tokens": 6,
"prefill": [
{
"id": 1,
"logprob": null,
"text": "<s>"
},
{
"id": 3735,
"logprob": -10.5,
"text": "Test"
},
{
"id": 2159,
"logprob": -12.140625,
"text": "request"
}
],
"prefill": [],
"seed": 0,
"tokens": [
{
......
This source diff could not be displayed because it is too large. You can view the blob instead.
......@@ -3,48 +3,7 @@
"best_of_sequences": null,
"finish_reason": "length",
"generated_tokens": 10,
"prefill": [
{
"id": 50278,
"logprob": null,
"text": "<|USER|>"
},
{
"id": 1276,
"logprob": -4.5546875,
"text": "What"
},
{
"id": 434,
"logprob": -4.1992188,
"text": "'s"
},
{
"id": 634,
"logprob": -5.125,
"text": " your"
},
{
"id": 12315,
"logprob": -9.8984375,
"text": " mood"
},
{
"id": 3063,
"logprob": -4.0976562,
"text": " today"
},
{
"id": 32,
"logprob": -0.14562988,
"text": "?"
},
{
"id": 50279,
"logprob": -0.26733398,
"text": "<|ASSISTANT|>"
}
],
"prefill": [],
"seed": null,
"tokens": [
{
......
......@@ -4,48 +4,7 @@
"best_of_sequences": null,
"finish_reason": "length",
"generated_tokens": 10,
"prefill": [
{
"id": 50278,
"logprob": null,
"text": "<|USER|>"
},
{
"id": 1276,
"logprob": -4.5546875,
"text": "What"
},
{
"id": 434,
"logprob": -4.1953125,
"text": "'s"
},
{
"id": 634,
"logprob": -5.125,
"text": " your"
},
{
"id": 12315,
"logprob": -9.8828125,
"text": " mood"
},
{
"id": 3063,
"logprob": -3.9980469,
"text": " today"
},
{
"id": 32,
"logprob": -0.14672852,
"text": "?"
},
{
"id": 50279,
"logprob": -0.26489258,
"text": "<|ASSISTANT|>"
}
],
"prefill": [],
"seed": null,
"tokens": [
{
......@@ -117,48 +76,7 @@
"best_of_sequences": null,
"finish_reason": "length",
"generated_tokens": 10,
"prefill": [
{
"id": 50278,
"logprob": null,
"text": "<|USER|>"
},
{
"id": 1276,
"logprob": -4.5546875,
"text": "What"
},
{
"id": 434,
"logprob": -4.1953125,
"text": "'s"
},
{
"id": 634,
"logprob": -5.125,
"text": " your"
},
{
"id": 12315,
"logprob": -9.8828125,
"text": " mood"
},
{
"id": 3063,
"logprob": -3.9980469,
"text": " today"
},
{
"id": 32,
"logprob": -0.14672852,
"text": "?"
},
{
"id": 50279,
"logprob": -0.26489258,
"text": "<|ASSISTANT|>"
}
],
"prefill": [],
"seed": null,
"tokens": [
{
......@@ -230,48 +148,7 @@
"best_of_sequences": null,
"finish_reason": "length",
"generated_tokens": 10,
"prefill": [
{
"id": 50278,
"logprob": null,
"text": "<|USER|>"
},
{
"id": 1276,
"logprob": -4.5546875,
"text": "What"
},
{
"id": 434,
"logprob": -4.1953125,
"text": "'s"
},
{
"id": 634,
"logprob": -5.125,
"text": " your"
},
{
"id": 12315,
"logprob": -9.8828125,
"text": " mood"
},
{
"id": 3063,
"logprob": -3.9980469,
"text": " today"
},
{
"id": 32,
"logprob": -0.14672852,
"text": "?"
},
{
"id": 50279,
"logprob": -0.26489258,
"text": "<|ASSISTANT|>"
}
],
"prefill": [],
"seed": null,
"tokens": [
{
......@@ -343,48 +220,7 @@
"best_of_sequences": null,
"finish_reason": "length",
"generated_tokens": 10,
"prefill": [
{
"id": 50278,
"logprob": null,
"text": "<|USER|>"
},
{
"id": 1276,
"logprob": -4.5546875,
"text": "What"
},
{
"id": 434,
"logprob": -4.1953125,
"text": "'s"
},
{
"id": 634,
"logprob": -5.125,
"text": " your"
},
{
"id": 12315,
"logprob": -9.8828125,
"text": " mood"
},
{
"id": 3063,
"logprob": -3.9980469,
"text": " today"
},
{
"id": 32,
"logprob": -0.14672852,
"text": "?"
},
{
"id": 50279,
"logprob": -0.26489258,
"text": "<|ASSISTANT|>"
}
],
"prefill": [],
"seed": null,
"tokens": [
{
......
......@@ -3,98 +3,7 @@
"best_of_sequences": null,
"finish_reason": "length",
"generated_tokens": 10,
"prefill": [
{
"id": 50278,
"logprob": null,
"text": "<|prompter|>"
},
{
"id": 1276,
"logprob": -8.0234375,
"text": "What"
},
{
"id": 310,
"logprob": -5.4179688,
"text": " is"
},
{
"id": 247,
"logprob": -2.1542969,
"text": " a"
},
{
"id": 1167,
"logprob": -5.359375,
"text": " mem"
},
{
"id": 70,
"logprob": -0.006038666,
"text": "e"
},
{
"id": 13,
"logprob": -7.328125,
"text": ","
},
{
"id": 285,
"logprob": -0.3173828,
"text": " and"
},
{
"id": 752,
"logprob": -2.0625,
"text": " what"
},
{
"id": 434,
"logprob": -5.7734375,
"text": "'s"
},
{
"id": 253,
"logprob": -0.74072266,
"text": " the"
},
{
"id": 2892,
"logprob": -6.5898438,
"text": " history"
},
{
"id": 3212,
"logprob": -2.2949219,
"text": " behind"
},
{
"id": 436,
"logprob": -11.40625,
"text": " this"
},
{
"id": 3159,
"logprob": -2.1113281,
"text": " word"
},
{
"id": 32,
"logprob": -0.008056641,
"text": "?"
},
{
"id": 0,
"logprob": -2.3300781,
"text": "<|endoftext|>"
},
{
"id": 50281,
"logprob": -18.28125,
"text": "<|assistant|>"
}
],
"prefill": [],
"seed": null,
"tokens": [
{
......
......@@ -4,98 +4,7 @@
"best_of_sequences": null,
"finish_reason": "length",
"generated_tokens": 10,
"prefill": [
{
"id": 50278,
"logprob": null,
"text": "<|prompter|>"
},
{
"id": 1276,
"logprob": -8.0234375,
"text": "What"
},
{
"id": 310,
"logprob": -5.4179688,
"text": " is"
},
{
"id": 247,
"logprob": -2.1542969,
"text": " a"
},
{
"id": 1167,
"logprob": -5.359375,
"text": " mem"
},
{
"id": 70,
"logprob": -0.006038666,
"text": "e"
},
{
"id": 13,
"logprob": -7.328125,
"text": ","
},
{
"id": 285,
"logprob": -0.3173828,
"text": " and"
},
{
"id": 752,
"logprob": -2.0625,
"text": " what"
},
{
"id": 434,
"logprob": -5.7734375,
"text": "'s"
},
{
"id": 253,
"logprob": -0.74072266,
"text": " the"
},
{
"id": 2892,
"logprob": -6.5898438,
"text": " history"
},
{
"id": 3212,
"logprob": -2.2949219,
"text": " behind"
},
{
"id": 436,
"logprob": -11.40625,
"text": " this"
},
{
"id": 3159,
"logprob": -2.1113281,
"text": " word"
},
{
"id": 32,
"logprob": -0.008056641,
"text": "?"
},
{
"id": 0,
"logprob": -2.3300781,
"text": "<|endoftext|>"
},
{
"id": 50281,
"logprob": -18.28125,
"text": "<|assistant|>"
}
],
"prefill": [],
"seed": null,
"tokens": [
{
......@@ -167,98 +76,7 @@
"best_of_sequences": null,
"finish_reason": "length",
"generated_tokens": 10,
"prefill": [
{
"id": 50278,
"logprob": null,
"text": "<|prompter|>"
},
{
"id": 1276,
"logprob": -8.0234375,
"text": "What"
},
{
"id": 310,
"logprob": -5.421875,
"text": " is"
},
{
"id": 247,
"logprob": -2.1640625,
"text": " a"
},
{
"id": 1167,
"logprob": -5.40625,
"text": " mem"
},
{
"id": 70,
"logprob": -0.005420685,
"text": "e"
},
{
"id": 13,
"logprob": -7.2226562,
"text": ","
},
{
"id": 285,
"logprob": -0.26879883,
"text": " and"
},
{
"id": 752,
"logprob": -2.1992188,
"text": " what"
},
{
"id": 434,
"logprob": -5.46875,
"text": "'s"
},
{
"id": 253,
"logprob": -0.8017578,
"text": " the"
},
{
"id": 2892,
"logprob": -6.6796875,
"text": " history"
},
{
"id": 3212,
"logprob": -2.1972656,
"text": " behind"
},
{
"id": 436,
"logprob": -11.4453125,
"text": " this"
},
{
"id": 3159,
"logprob": -2.1933594,
"text": " word"
},
{
"id": 32,
"logprob": -0.007858276,
"text": "?"
},
{
"id": 0,
"logprob": -2.328125,
"text": "<|endoftext|>"
},
{
"id": 50281,
"logprob": -18.21875,
"text": "<|assistant|>"
}
],
"prefill": [],
"seed": null,
"tokens": [
{
......@@ -330,98 +148,7 @@
"best_of_sequences": null,
"finish_reason": "length",
"generated_tokens": 10,
"prefill": [
{
"id": 50278,
"logprob": null,
"text": "<|prompter|>"
},
{
"id": 1276,
"logprob": -8.0234375,
"text": "What"
},
{
"id": 310,
"logprob": -5.421875,
"text": " is"
},
{
"id": 247,
"logprob": -2.1640625,
"text": " a"
},
{
"id": 1167,
"logprob": -5.40625,
"text": " mem"
},
{
"id": 70,
"logprob": -0.005420685,
"text": "e"
},
{
"id": 13,
"logprob": -7.2226562,
"text": ","
},
{
"id": 285,
"logprob": -0.26879883,
"text": " and"
},
{
"id": 752,
"logprob": -2.1992188,
"text": " what"
},
{
"id": 434,
"logprob": -5.46875,
"text": "'s"
},
{
"id": 253,
"logprob": -0.8017578,
"text": " the"
},
{
"id": 2892,
"logprob": -6.6796875,
"text": " history"
},
{
"id": 3212,
"logprob": -2.1972656,
"text": " behind"
},
{
"id": 436,
"logprob": -11.4453125,
"text": " this"
},
{
"id": 3159,
"logprob": -2.1933594,
"text": " word"
},
{
"id": 32,
"logprob": -0.007858276,
"text": "?"
},
{
"id": 0,
"logprob": -2.328125,
"text": "<|endoftext|>"
},
{
"id": 50281,
"logprob": -18.21875,
"text": "<|assistant|>"
}
],
"prefill": [],
"seed": null,
"tokens": [
{
......@@ -493,98 +220,7 @@
"best_of_sequences": null,
"finish_reason": "length",
"generated_tokens": 10,
"prefill": [
{
"id": 50278,
"logprob": null,
"text": "<|prompter|>"
},
{
"id": 1276,
"logprob": -8.0234375,
"text": "What"
},
{
"id": 310,
"logprob": -5.421875,
"text": " is"
},
{
"id": 247,
"logprob": -2.1640625,
"text": " a"
},
{
"id": 1167,
"logprob": -5.40625,
"text": " mem"
},
{
"id": 70,
"logprob": -0.005420685,
"text": "e"
},
{
"id": 13,
"logprob": -7.2226562,
"text": ","
},
{
"id": 285,
"logprob": -0.26879883,
"text": " and"
},
{
"id": 752,
"logprob": -2.1992188,
"text": " what"
},
{
"id": 434,
"logprob": -5.46875,
"text": "'s"
},
{
"id": 253,
"logprob": -0.8017578,
"text": " the"
},
{
"id": 2892,
"logprob": -6.6796875,
"text": " history"
},
{
"id": 3212,
"logprob": -2.1972656,
"text": " behind"
},
{
"id": 436,
"logprob": -11.4453125,
"text": " this"
},
{
"id": 3159,
"logprob": -2.1933594,
"text": " word"
},
{
"id": 32,
"logprob": -0.007858276,
"text": "?"
},
{
"id": 0,
"logprob": -2.328125,
"text": "<|endoftext|>"
},
{
"id": 50281,
"logprob": -18.21875,
"text": "<|assistant|>"
}
],
"prefill": [],
"seed": null,
"tokens": [
{
......
......@@ -3,23 +3,7 @@
"best_of_sequences": null,
"finish_reason": "length",
"generated_tokens": 10,
"prefill": [
{
"id": 1,
"logprob": null,
"text": "<s>"
},
{
"id": 4321,
"logprob": -9.8359375,
"text": "Test"
},
{
"id": 2009,
"logprob": -9.6171875,
"text": "request"
}
],
"prefill": [],
"seed": null,
"tokens": [
{
......
......@@ -3,23 +3,7 @@
"best_of_sequences": null,
"finish_reason": "length",
"generated_tokens": 10,
"prefill": [
{
"id": 1,
"logprob": null,
"text": "<s>"
},
{
"id": 4321,
"logprob": -9.7890625,
"text": "Test"
},
{
"id": 2009,
"logprob": -9.625,
"text": "request"
}
],
"prefill": [],
"seed": 0,
"tokens": [
{
......
......@@ -4,23 +4,7 @@
"best_of_sequences": null,
"finish_reason": "length",
"generated_tokens": 10,
"prefill": [
{
"id": 1,
"logprob": null,
"text": "<s>"
},
{
"id": 4321,
"logprob": -9.8828125,
"text": "Test"
},
{
"id": 2009,
"logprob": -9.5859375,
"text": "request"
}
],
"prefill": [],
"seed": null,
"tokens": [
{
......@@ -93,23 +77,7 @@
"best_of_sequences": null,
"finish_reason": "length",
"generated_tokens": 10,
"prefill": [
{
"id": 1,
"logprob": null,
"text": "<s>"
},
{
"id": 4321,
"logprob": -9.796875,
"text": "Test"
},
{
"id": 2009,
"logprob": -9.625,
"text": "request"
}
],
"prefill": [],
"seed": null,
"tokens": [
{
......@@ -182,23 +150,7 @@
"best_of_sequences": null,
"finish_reason": "length",
"generated_tokens": 10,
"prefill": [
{
"id": 1,
"logprob": null,
"text": "<s>"
},
{
"id": 4321,
"logprob": -9.9296875,
"text": "Test"
},
{
"id": 2009,
"logprob": -9.5703125,
"text": "request"
}
],
"prefill": [],
"seed": null,
"tokens": [
{
......@@ -271,23 +223,7 @@
"best_of_sequences": null,
"finish_reason": "length",
"generated_tokens": 10,
"prefill": [
{
"id": 1,
"logprob": null,
"text": "<s>"
},
{
"id": 4321,
"logprob": -9.84375,
"text": "Test"
},
{
"id": 2009,
"logprob": -9.6171875,
"text": "request"
}
],
"prefill": [],
"seed": null,
"tokens": [
{
......
......@@ -6,6 +6,7 @@ def flash_phi35_moe_handle(launcher):
with launcher(
"microsoft/Phi-3.5-MoE-instruct",
num_shard=4,
max_batch_prefill_tokens=10000,
) as handle:
yield handle
......
......@@ -4,7 +4,10 @@ import asyncio
@pytest.fixture(scope="module")
def mllama_handle(launcher):
with launcher("meta-llama/Llama-3.2-11B-Vision-Instruct", num_shard=2) as handle:
with launcher(
"meta-llama/Llama-3.2-11B-Vision-Instruct",
num_shard=2,
) as handle:
yield handle
......@@ -75,7 +78,9 @@ async def test_mllama_load(mllama, generate_load, response_snapshot):
},
],
)
for i in range(4)
# TODO with v3, 4 breaks here. Nothing accounts of the image VRAM
# because mllama is the only one doing its thing.
for i in range(2)
]
responses = await asyncio.gather(*futures)
......
......@@ -29,6 +29,26 @@ use tracing_subscriber::{filter::LevelFilter, EnvFilter};
mod env_runtime;
mod gpu;
fn compute_optimal(config: Option<&Config>, compute: Option<&ComputeType>) -> Option<usize> {
if let (Some(config), Some(compute)) = (config, compute) {
if let (Some(f16_max_compute), Some(model_compute)) = (compute.f16_flop(), config.flop()) {
tracing::debug!("MAx compute {f16_max_compute} model compute {model_compute}");
let optimal_size = (f16_max_compute / model_compute) as usize;
if optimal_size > 100 {
// Ignore calculations that's too low
// Most likely an error
Some(optimal_size)
} else {
None
}
} else {
None
}
} else {
None
}
}
fn get_config(
model_id: &str,
revision: &Option<String>,
......@@ -144,10 +164,15 @@ struct RawConfig {
quantization_config: Option<QuantizationConfig>,
n_embd: Option<usize>,
hidden_size: Option<usize>,
intermediate_size: Option<usize>,
num_attention_heads: Option<usize>,
num_key_value_heads: Option<usize>,
num_hidden_layers: Option<usize>,
head_dim: Option<usize>,
vision_config: Option<VisionConfig>,
is_encoder_decoder: Option<bool>,
#[serde(rename = "num_experts_per_tok")]
experts: Option<usize>,
}
#[derive(Deserialize)]
......@@ -155,17 +180,56 @@ struct QuantizationConfig {
quant_method: Option<Quantization>,
}
#[derive(Deserialize)]
#[derive(Debug, Deserialize)]
struct VisionConfig {}
#[derive(Deserialize)]
#[derive(Debug, Deserialize)]
struct Config {
max_position_embeddings: Option<usize>,
quantize: Option<Quantization>,
head_dim: Option<usize>,
num_heads: Option<usize>,
num_kv_heads: Option<usize>,
num_layers: Option<usize>,
intermediate_size: Option<usize>,
hidden_size: Option<usize>,
model_type: Option<String>,
vision_config: Option<VisionConfig>,
is_encoder_decoder: bool,
experts: Option<usize>,
}
impl Config {
fn flop(&self) -> Option<u64> {
if self.vision_config.is_some() {
// VLM are much harder to predict and VRAM requirements
// Are more complex.
return None;
}
let num_heads = self.num_heads? as u64;
let num_kv_heads = self.num_kv_heads? as u64;
let head_dim = self.head_dim? as u64;
let hidden_size = self.hidden_size? as u64;
let intermediate_size = if let Some(experts) = self.experts {
(self.intermediate_size? * experts) as u64
} else {
self.intermediate_size? as u64
};
let num_layers = self.num_layers? as u64;
let q_flops = 2 * num_heads * head_dim * hidden_size;
let k_flops = 2 * num_kv_heads * head_dim * hidden_size;
let v_flops = 2 * num_kv_heads * head_dim * hidden_size;
let attn_flops = 2 * num_heads * head_dim * hidden_size;
let o_flops = 2 * num_heads * head_dim * hidden_size;
let attn_layer_flops = q_flops + k_flops + v_flops + attn_flops + o_flops;
let gate_up_down_flops = 2 * 3 * hidden_size * intermediate_size;
let layer_flops = attn_layer_flops + gate_up_down_flops;
let total = layer_flops * num_layers;
Some(total)
}
}
impl From<RawConfig> for Config {
......@@ -175,25 +239,25 @@ impl From<RawConfig> for Config {
.or(other.max_seq_len)
.or(other.n_positions);
let quantize = other.quantization_config.and_then(|q| q.quant_method);
let head_dim = other.head_dim.or_else(|| {
match (other.hidden_size, other.n_embd, other.num_attention_heads) {
(Some(hidden_size), _, Some(num_attention_heads))
if hidden_size % num_attention_heads == 0 =>
{
Some(hidden_size / num_attention_heads)
}
// Legacy
(_, Some(hidden_size), Some(num_attention_heads))
let hidden_size = other.hidden_size.or(other.n_embd);
let head_dim = other
.head_dim
.or_else(|| match (hidden_size, other.num_attention_heads) {
(Some(hidden_size), Some(num_attention_heads))
if hidden_size % num_attention_heads == 0 =>
{
Some(hidden_size / num_attention_heads)
}
_ => None,
}
});
});
let num_heads = other.num_attention_heads;
let num_layers = other.num_hidden_layers;
let num_kv_heads = other.num_key_value_heads.or(other.num_attention_heads);
let intermediate_size = other.intermediate_size;
let model_type = other.model_type;
let vision_config = other.vision_config;
let is_encoder_decoder = other.is_encoder_decoder.unwrap_or(false);
let experts = other.experts;
Config {
max_position_embeddings,
quantize,
......@@ -201,6 +265,12 @@ impl From<RawConfig> for Config {
model_type,
vision_config,
is_encoder_decoder,
hidden_size,
num_heads,
num_kv_heads,
intermediate_size,
num_layers,
experts,
}
}
}
......@@ -698,6 +768,14 @@ struct Args {
/// Default is 2MB
#[clap(default_value = "2000000", long, env)]
payload_limit: usize,
/// Enables prefill logprobs
///
/// Logprobs in the prompt are deactivated by default because they consume
/// a large amount of VRAM (especially for long prompts).
/// Using this flag reallows users to ask for them.
#[clap(long, env)]
enable_prefill_logprobs: bool,
}
#[derive(Debug)]
......@@ -733,6 +811,7 @@ fn shard_manager(
max_batch_size: Option<usize>,
max_input_tokens: Option<usize>,
lora_adapters: Option<String>,
enable_prefill_logprobs: bool,
otlp_endpoint: Option<String>,
otlp_service_name: String,
log_level: LevelFilter,
......@@ -882,6 +961,11 @@ fn shard_manager(
envs.push(("LORA_ADAPTERS".into(), lora_adapters.into()));
}
// Logprobs
if enable_prefill_logprobs {
envs.push(("REQUEST_LOGPROBS".into(), "1".into()));
}
// If huggingface_hub_cache is some, pass it to the shard
// Useful when running inside a docker container
if let Some(huggingface_hub_cache) = huggingface_hub_cache {
......@@ -1373,6 +1457,7 @@ fn spawn_shards(
let rope_factor = args.rope_factor;
let max_batch_size = args.max_batch_size;
let lora_adapters = args.lora_adapters.clone();
let enable_prefill_logprobs = args.enable_prefill_logprobs;
thread::spawn(move || {
shard_manager(
model_id,
......@@ -1400,6 +1485,7 @@ fn spawn_shards(
max_batch_size,
max_input_tokens,
lora_adapters,
enable_prefill_logprobs,
otlp_endpoint,
otlp_service_name,
max_log_level,
......@@ -1439,7 +1525,45 @@ fn spawn_shards(
Ok(())
}
fn compute_type(num_shard: usize) -> Option<String> {
#[derive(Debug)]
struct ComputeType {
count: usize,
card: String,
}
impl ComputeType {
fn f16_flop(&self) -> Option<u64> {
let card_flop = match &self.card[..] {
// https://www.nvidia.com/en-us/geforce/graphics-cards/40-series/rtx-4090/
// Specs are unclear https://www.itcreations.com/nvidia-gpu/nvidia-geforce-rtx-4090-gpu
"nvidia-4090" => Some(82 * 10u64.pow(12)),
// https://www.nvidia.com/en-us/data-center/tesla-t4/
"nvidia-t4" => Some(65 * 10u64.pow(12)),
// https://www.nvidia.com/en-us/data-center/l4/
"nvidia-l4" => Some(121 * 10u64.pow(12)),
// https://www.nvidia.com/en-us/data-center/products/a10-gpu/
"nvidia-a10g" => Some(125 * 10u64.pow(12)),
// https://www.nvidia.com/en-us/data-center/h100/
// https://www.techpowerup.com/gpu-specs/docs/nvidia-gh100-architecture.pdf
"nvidia-h100-80gb-hbm3" => Some(900 * 10u64.pow(12)),
// https://www.nvidia.com/content/dam/en-zz/Solutions/Data-Center/a100/pdf/nvidia-a100-datasheet-us-nvidia-1758950-r4-web.pdf
"nvidia-a100" => Some(312 * 10u64.pow(12)),
card => {
tracing::warn!("Unkown compute for card {card}");
None
}
};
card_flop.map(|f| f * self.count as u64)
}
}
impl From<ComputeType> for OsString {
fn from(value: ComputeType) -> Self {
format!("{}-{}", value.count, value.card).into()
}
}
fn compute_type(num_shard: usize) -> Option<ComputeType> {
let output = Command::new("nvidia-smi")
.args(["--query-gpu=gpu_name", "--format=csv"])
.output()
......@@ -1447,8 +1571,10 @@ fn compute_type(num_shard: usize) -> Option<String> {
let output = String::from_utf8(output.stdout).ok()?;
let fullname = output.split('\n').nth(1)?;
let cardname = fullname.replace(' ', "-").to_lowercase();
let compute_type = format!("{num_shard}-{cardname}");
Some(compute_type)
Some(ComputeType {
count: num_shard,
card: cardname,
})
}
fn spawn_webserver(
......@@ -1700,26 +1826,22 @@ fn main() -> Result<(), LauncherError> {
let config: Option<Config> = get_config(&args.model_id, &args.revision).ok();
let quantize = config.as_ref().and_then(|c| c.quantize);
// Quantization usually means you're even more RAM constrained.
let max_default = 4096;
let max_position_embeddings = if let Some(config) = &config {
if let Some(max_position_embeddings) = config.max_position_embeddings {
if max_position_embeddings > max_default {
max_default
} else {
max_position_embeddings
}
} else {
max_default
}
} else {
max_default
};
let (prefix_caching, attention) = resolve_attention(&config, &args.lora_adapters);
tracing::info!("Using attention {attention} - Prefix caching {prefix_caching}");
std::env::set_var("PREFIX_CACHING", prefix_caching);
std::env::set_var("ATTENTION", attention);
let num_shard = find_num_shards(args.sharded, args.num_shard)?;
if num_shard > 1 {
if matches!(args.quantize, Some(Quantization::Exl2)) {
return Err(LauncherError::ArgumentValidation(
"Sharding is currently not supported with `exl2` quantization".into(),
));
}
tracing::info!("Sharding model on {num_shard} processes");
}
let max_input_tokens = {
match (args.max_input_tokens, args.max_input_length) {
(Some(max_input_tokens), Some(max_input_length)) => {
......@@ -1739,9 +1861,17 @@ fn main() -> Result<(), LauncherError> {
Some(max_batch_prefill_tokens) => max_batch_prefill_tokens,
None => {
// TODO figure out hardware optimal value
let value = 4096.min(max_position_embeddings as u32);
let compute_type = compute_type(num_shard);
let compute_optimal = compute_optimal(config.as_ref(), compute_type.as_ref());
let default = compute_optimal.unwrap_or(4096);
let max_position_embeddings = config.and_then(|c| c.max_position_embeddings);
let value = if let Some(max_position_embeddings) = max_position_embeddings {
default.min(max_position_embeddings)
} else {
default
};
tracing::info!("Default `max_batch_prefill_tokens` to {value}");
value
value as u32
}
}
};
......@@ -1796,16 +1926,6 @@ fn main() -> Result<(), LauncherError> {
);
}
let num_shard = find_num_shards(args.sharded, args.num_shard)?;
if num_shard > 1 {
if matches!(args.quantize, Some(Quantization::Exl2)) {
return Err(LauncherError::ArgumentValidation(
"Sharding is currently not supported with `exl2` quantization".into(),
));
}
tracing::info!("Sharding model on {num_shard} processes");
}
if let Some(ref max_batch_total_tokens) = args.max_batch_total_tokens {
if let Some(max_total_tokens) = max_total_tokens {
if max_total_tokens as u32 > *max_batch_total_tokens {
......
ShareGPT_V3_unfiltered_cleaned_split.json:
wget https://huggingface.co/datasets/anon8231489123/ShareGPT_Vicuna_unfiltered/resolve/main/ShareGPT_V3_unfiltered_cleaned_split.json
prepare_share: ShareGPT_V3_unfiltered_cleaned_split.json
python filter.py
prepare_orca:
python orca.py
import { check } from 'k6';
import { scenario } from 'k6/execution';
import http from 'k6/http';
import { Trend, Counter } from 'k6/metrics';
const host = __ENV.HOST;
const model_id = __ENV.MODEL_ID;
const timePerToken = new Trend('time_per_token', true);
const tokens = new Counter('tokens');
const new_tokens = new Counter('new_tokens');
const input_tokens = new Counter('input_tokens');
const max_new_tokens = 50;
// const shareGPT = JSON.parse(open("ShareGPT_V3_unfiltered_cleaned_split.json"))
const shareGPT = JSON.parse(open("small.json"))
export function get_options() {
return {
thresholds: {
http_req_failed: ['rate==0'],
// time_per_token: [{
// threshold: `p(50)<${5 * reference_latency_ms}`,
// abortOnFail: true,
// delayAbortEval: '10s'
// }],
},
scenarios: {
// single_user: {
// executor: 'constant-arrival-rate',
// duration: '60s',
// preAllocatedVUs: 1,
// rate: 20,
// timeUnit: '1s',
// },
// load_test: {
// executor: 'constant-arrival-rate',
// duration: '60s',
// preAllocatedVUs: 100,
// rate: 1,
// timeUnit: '1s',
// },
// breakpoint: {
// executor: 'ramping-arrival-rate', //Assure load increase if the system slows
// preAllocatedVUs: 300,
// stages: [
// { duration: '60s', target: 100 }, // just slowly ramp-up to a HUGE load
// ],
// },
throughput: {
executor: 'shared-iterations',
vus: 100,
iterations: 200,
maxDuration: '40s',
},
},
};
}
function generate_payload(gpt, max_new_tokens) {
const input = gpt["conversations"][0]["value"];
return { "messages": [{ "role": "user", "content": input }], "temperature": 0, "model": `${model_id}`, "max_tokens": max_new_tokens }
}
export const options = get_options();
export default function run() {
const headers = { 'Content-Type': 'application/json' };
const query = shareGPT[scenario.iterationInTest % shareGPT.length];
const payload = JSON.stringify(generate_payload(query, max_new_tokens));
const res = http.post(`http://${host}/v1/chat/completions`, payload, {
headers,
});
if (res.status >= 400 && res.status < 500) {
return;
}
check(res, {
'Post status is 200': (res) => res.status === 200,
});
const duration = res.timings.duration;
if (res.status === 200) {
const body = res.json();
const completion_tokens = body.usage.completion_tokens;
const latency_ms_per_token = duration / completion_tokens;
timePerToken.add(latency_ms_per_token);
const prompt_tokens = body.usage.prompt_tokens;
input_tokens.add(prompt_tokens);
new_tokens.add(completion_tokens);
tokens.add(completion_tokens + prompt_tokens);
}
}
import json
def main():
with open("./ShareGPT_V3_unfiltered_cleaned_split.json", "r") as f:
data = json.load(f)
# Select only the first 2k conversations that start with a human.
max = 2000
conversations = []
for conversation in data:
conv = conversation.get("conversations")
if conv and conv[0]["from"] == "human":
# Trim the rest of the output
conversation["conversations"] = conversation["conversations"][:1]
conversations.append(conversation)
if len(conversation) >= max:
break
with open("./small.json", "w") as f:
data = json.dump(conversations, f, indent=4)
if __name__ == "__main__":
main()
Markdown is supported
0% or .
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment