Skip to content
GitLab
Menu
Projects
Groups
Snippets
Loading...
Help
Help
Support
Community forum
Keyboard shortcuts
?
Submit feedback
Contribute to GitLab
Sign in / Register
Toggle navigation
Menu
Open sidebar
OpenDAS
text-generation-inference
Commits
87dc034b
Unverified
Commit
87dc034b
authored
May 31, 2023
by
OlivierDehaene
Committed by
GitHub
May 31, 2023
Browse files
feat(server): add retry on download (#384)
parent
444400b4
Changes
4
Hide whitespace changes
Inline
Side-by-side
Showing
4 changed files
with
309 additions
and
298 deletions
+309
-298
integration-tests/models/__snapshots__/test_flash_falcon/test_flash_falcon_load.json
...snapshots__/test_flash_falcon/test_flash_falcon_load.json
+281
-281
server/text_generation_server/models/flash_rw.py
server/text_generation_server/models/flash_rw.py
+2
-2
server/text_generation_server/models/rw.py
server/text_generation_server/models/rw.py
+1
-1
server/text_generation_server/utils/hub.py
server/text_generation_server/utils/hub.py
+25
-14
No files found.
integration-tests/models/__snapshots__/test_flash_falcon/test_flash_falcon_load.json
View file @
87dc034b
...
...
@@ -12,22 +12,22 @@
},
{
"id"
:
330
,
"logprob"
:
-5.
71
875
,
"logprob"
:
-5.
96
875
,
"text"
:
"ir"
},
{
"id"
:
1622
,
"logprob"
:
-5.6
2
5
,
"logprob"
:
-5.6
17187
5
,
"text"
:
"af"
},
{
"id"
:
249
,
"logprob"
:
-6.5
3125
,
"logprob"
:
-6.5
039062
,
"text"
:
"at"
},
{
"id"
:
1480
,
"logprob"
:
-8.0
6
25
,
"logprob"
:
-8.0
7031
25
,
"text"
:
"ron"
},
{
...
...
@@ -37,27 +37,27 @@
},
{
"id"
:
23866
,
"logprob"
:
-9.
62
5
,
"logprob"
:
-9.
5937
5
,
"text"
:
" obsessed"
},
{
"id"
:
335
,
"logprob"
:
-0.048
583984
,
"logprob"
:
-0.048
37036
,
"text"
:
" with"
},
{
"id"
:
26680
,
"logprob"
:
-3.9
84375
,
"logprob"
:
-3.9
960938
,
"text"
:
" gir"
},
{
"id"
:
1903
,
"logprob"
:
-0.07
617187
5
,
"logprob"
:
-0.07
52563
5
,
"text"
:
"aff"
},
{
"id"
:
255
,
"logprob"
:
-0.006
6833496
,
"logprob"
:
-0.006
790161
,
"text"
:
"es"
},
{
...
...
@@ -67,82 +67,82 @@
},
{
"id"
:
248
,
"logprob"
:
-4.3
4375
,
"logprob"
:
-4.3
320312
,
"text"
:
" the"
},
{
"id"
:
758
,
"logprob"
:
-3.73
4375
,
"logprob"
:
-3.73
63281
,
"text"
:
" most"
},
{
"id"
:
21735
,
"logprob"
:
-5.1
2
5
,
"logprob"
:
-5.1
0937
5
,
"text"
:
" glorious"
},
{
"id"
:
5985
,
"logprob"
:
-2.0
7812
5
,
"logprob"
:
-2.0
937
5
,
"text"
:
" animal"
},
{
"id"
:
313
,
"logprob"
:
-1.1
953125
,
"logprob"
:
-1.1
845703
,
"text"
:
" on"
},
{
"id"
:
248
,
"logprob"
:
-0.7
812
5
,
"logprob"
:
-0.7
773437
5
,
"text"
:
" the"
},
{
"id"
:
1936
,
"logprob"
:
-2.3
906
25
,
"logprob"
:
-2.3
8281
25
,
"text"
:
" face"
},
{
"id"
:
275
,
"logprob"
:
-0.0044
555664
,
"logprob"
:
-0.0044
403076
,
"text"
:
" of"
},
{
"id"
:
414
,
"logprob"
:
-1.9
84375
,
"logprob"
:
-1.9
667969
,
"text"
:
" this"
},
{
"id"
:
6490
,
"logprob"
:
-2.0
3125
,
"logprob"
:
-2.0
449219
,
"text"
:
" Earth"
},
{
"id"
:
25
,
"logprob"
:
-0.28
32031
2
,
"logprob"
:
-0.28
19824
2
,
"text"
:
"."
},
{
"id"
:
401
,
"logprob"
:
-7.9
062
5
,
"logprob"
:
-7.9
2187
5
,
"text"
:
" G"
},
{
"id"
:
6013
,
"logprob"
:
-2.2
65625
,
"logprob"
:
-2.2
714844
,
"text"
:
"ira"
},
{
"id"
:
694
,
"logprob"
:
-0.6
40625
,
"logprob"
:
-0.6
2353516
,
"text"
:
"ft"
},
{
"id"
:
1480
,
"logprob"
:
-0.20
3125
,
"logprob"
:
-0.20
947266
,
"text"
:
"ron"
},
{
"id"
:
9369
,
"logprob"
:
-4.5
3
12
5
,
"logprob"
:
-4.5
5078
12
,
"text"
:
" believes"
},
{
...
...
@@ -152,32 +152,32 @@
},
{
"id"
:
599
,
"logprob"
:
-2.7
5
,
"logprob"
:
-2.7
402344
,
"text"
:
" other"
},
{
"id"
:
5632
,
"logprob"
:
-0.218
75
,
"logprob"
:
-0.218
99414
,
"text"
:
" animals"
},
{
"id"
:
362
,
"logprob"
:
-0.76
171875
,
"logprob"
:
-0.76
708984
,
"text"
:
" are"
},
{
"id"
:
23981
,
"logprob"
:
-4.9
6875
,
"logprob"
:
-4.9
960938
,
"text"
:
" irrelevant"
},
{
"id"
:
635
,
"logprob"
:
-4.2
18
75
,
"logprob"
:
-4.2
343
75
,
"text"
:
" when"
},
{
"id"
:
4354
,
"logprob"
:
-0.51
953125
,
"logprob"
:
-0.51
31836
,
"text"
:
" compared"
},
{
...
...
@@ -187,47 +187,47 @@
},
{
"id"
:
248
,
"logprob"
:
-0.58
984375
,
"logprob"
:
-0.58
447266
,
"text"
:
" the"
},
{
"id"
:
21735
,
"logprob"
:
-3.6875
,
"logprob"
:
-3.
679
6875
,
"text"
:
" glorious"
},
{
"id"
:
64398
,
"logprob"
:
-1.8
359375
,
"logprob"
:
-1.8
222656
,
"text"
:
" majesty"
},
{
"id"
:
275
,
"logprob"
:
-0.2
4316406
,
"logprob"
:
-0.2
3583984
,
"text"
:
" of"
},
{
"id"
:
248
,
"logprob"
:
-0.35
15625
,
"logprob"
:
-0.35
44922
,
"text"
:
" the"
},
{
"id"
:
26680
,
"logprob"
:
-0.24
414062
,
"logprob"
:
-0.24
609375
,
"text"
:
" gir"
},
{
"id"
:
23226
,
"logprob"
:
-0.0
3100586
,
"logprob"
:
-0.0
2960205
,
"text"
:
"affe"
},
{
"id"
:
25
,
"logprob"
:
-0.173
82812
,
"logprob"
:
-0.173
58398
,
"text"
:
"."
},
{
"id"
:
193
,
"logprob"
:
-1.39
84375
,
"logprob"
:
-1.39
25781
,
"text"
:
"
\n
"
},
{
...
...
@@ -237,7 +237,7 @@
},
{
"id"
:
37
,
"logprob"
:
-4.5
9375
,
"logprob"
:
-4.5
898438
,
"text"
:
":"
},
{
...
...
@@ -247,67 +247,67 @@
},
{
"id"
:
23
,
"logprob"
:
-
1.0
,
"logprob"
:
-
0.99365234
,
"text"
:
","
},
{
"id"
:
29033
,
"logprob"
:
-2.2
1875
,
"logprob"
:
-2.2
304688
,
"text"
:
" Gir"
},
{
"id"
:
1622
,
"logprob"
:
-0.10
644531
,
"logprob"
:
-0.10
7788086
,
"text"
:
"af"
},
{
"id"
:
249
,
"logprob"
:
-0.04
1992188
,
"logprob"
:
-0.04
257202
,
"text"
:
"at"
},
{
"id"
:
1480
,
"logprob"
:
-0.002
5024414
,
"logprob"
:
-0.002
4871826
,
"text"
:
"ron"
},
{
"id"
:
12
,
"logprob"
:
-1.42
96875
,
"logprob"
:
-1.42
77344
,
"text"
:
"!"
},
{
"id"
:
193
,
"logprob"
:
-1.10
15625
,
"logprob"
:
-1.10
05859
,
"text"
:
"
\n
"
},
{
"id"
:
50
,
"logprob"
:
-0.05
810547
,
"logprob"
:
-0.05
6915283
,
"text"
:
"G"
},
{
"id"
:
330
,
"logprob"
:
-0.1
2597656
,
"logprob"
:
-0.1
315918
,
"text"
:
"ir"
},
{
"id"
:
1622
,
"logprob"
:
-0.007
080078
,
"logprob"
:
-0.007
1105957
,
"text"
:
"af"
},
{
"id"
:
249
,
"logprob"
:
-0.008
300781
,
"logprob"
:
-0.008
453369
,
"text"
:
"at"
},
{
"id"
:
1480
,
"logprob"
:
-0.0006
866455
,
"logprob"
:
-0.0006
928444
,
"text"
:
"ron"
},
{
"id"
:
37
,
"logprob"
:
-0.0074
157715
,
"logprob"
:
-0.0074
920654
,
"text"
:
":"
}
],
...
...
@@ -315,37 +315,37 @@
"tokens"
:
[
{
"id"
:
23090
,
"logprob"
:
-1.82
03
125
,
"logprob"
:
-1.82
8
125
,
"special"
:
false
,
"text"
:
" Hello"
},
{
"id"
:
23
,
"logprob"
:
-0.3
2226562
,
"logprob"
:
-0.3
178711
,
"special"
:
false
,
"text"
:
","
},
{
"id"
:
8156
,
"logprob"
:
-0.23
828125
,
"logprob"
:
-0.23
925781
,
"special"
:
false
,
"text"
:
" Daniel"
},
{
"id"
:
12
,
"logprob"
:
-0.5
859375
,
"logprob"
:
-0.5
698242
,
"special"
:
false
,
"text"
:
"!"
},
{
"id"
:
193
,
"logprob"
:
-0.61
71875
,
"logprob"
:
-0.61
279297
,
"special"
:
false
,
"text"
:
"
\n
"
},
{
"id"
:
23626
,
"logprob"
:
-0.
39648438
,
"logprob"
:
-0.
4177246
,
"special"
:
false
,
"text"
:
"Daniel"
},
...
...
@@ -356,26 +356,26 @@
"text"
:
":"
},
{
"id"
:
295
,
"logprob"
:
-2.0
78125
,
"id"
:
1634
,
"logprob"
:
-2.0
605469
,
"special"
:
false
,
"text"
:
"
I
"
"text"
:
"
What
"
},
{
"id"
:
18
,
"logprob"
:
-1.
453125
,
"logprob"
:
-1.
5283203
,
"special"
:
false
,
"text"
:
"'"
},
{
"id"
:
88
,
"logprob"
:
-0.
4746093
8
,
"id"
:
94
,
"logprob"
:
-0.
00796508
8
,
"special"
:
false
,
"text"
:
"
m
"
"text"
:
"
s
"
}
]
},
"generated_text"
:
" Hello, Daniel!
\n
Daniel:
I'm
"
"generated_text"
:
" Hello, Daniel!
\n
Daniel:
What's
"
},
{
"details"
:
{
...
...
@@ -390,22 +390,22 @@
},
{
"id"
:
330
,
"logprob"
:
-5.
71
875
,
"logprob"
:
-5.
96
875
,
"text"
:
"ir"
},
{
"id"
:
1622
,
"logprob"
:
-5.6
2
5
,
"logprob"
:
-5.6
17187
5
,
"text"
:
"af"
},
{
"id"
:
249
,
"logprob"
:
-6.5
3125
,
"logprob"
:
-6.5
,
"text"
:
"at"
},
{
"id"
:
1480
,
"logprob"
:
-8.0
6
25
,
"logprob"
:
-8.0
7031
25
,
"text"
:
"ron"
},
{
...
...
@@ -415,197 +415,197 @@
},
{
"id"
:
23866
,
"logprob"
:
-9.
62
5
,
"logprob"
:
-9.
5937
5
,
"text"
:
" obsessed"
},
{
"id"
:
335
,
"logprob"
:
-0.048
58
3984
,
"logprob"
:
-0.048
3
3984
4
,
"text"
:
" with"
},
{
"id"
:
26680
,
"logprob"
:
-
3.984375
,
"logprob"
:
-
4.0
,
"text"
:
" gir"
},
{
"id"
:
1903
,
"logprob"
:
-0.07
6171875
,
"logprob"
:
-0.07
531738
,
"text"
:
"aff"
},
{
"id"
:
255
,
"logprob"
:
-0.006
683349
6
,
"logprob"
:
-0.006
79397
6
,
"text"
:
"es"
},
{
"id"
:
23
,
"logprob"
:
-1.54
6875
,
"logprob"
:
-1.54
78516
,
"text"
:
","
},
{
"id"
:
248
,
"logprob"
:
-4.3
4375
,
"logprob"
:
-4.3
320312
,
"text"
:
" the"
},
{
"id"
:
758
,
"logprob"
:
-3.73
4375
,
"logprob"
:
-3.73
63281
,
"text"
:
" most"
},
{
"id"
:
21735
,
"logprob"
:
-5.12
5
,
"logprob"
:
-5.1
13281
2
,
"text"
:
" glorious"
},
{
"id"
:
5985
,
"logprob"
:
-2.0
78125
,
"logprob"
:
-2.0
957031
,
"text"
:
" animal"
},
{
"id"
:
313
,
"logprob"
:
-1.1
953125
,
"logprob"
:
-1.1
835938
,
"text"
:
" on"
},
{
"id"
:
248
,
"logprob"
:
-0.7
8125
,
"logprob"
:
-0.7
7685547
,
"text"
:
" the"
},
{
"id"
:
1936
,
"logprob"
:
-2.3
90625
,
"logprob"
:
-2.3
808594
,
"text"
:
" face"
},
{
"id"
:
275
,
"logprob"
:
-0.0044
555664
,
"logprob"
:
-0.0044
36493
,
"text"
:
" of"
},
{
"id"
:
414
,
"logprob"
:
-1.9
84375
,
"logprob"
:
-1.9
638672
,
"text"
:
" this"
},
{
"id"
:
6490
,
"logprob"
:
-2.0
3125
,
"logprob"
:
-2.0
449219
,
"text"
:
" Earth"
},
{
"id"
:
25
,
"logprob"
:
-0.28
32031
2
,
"logprob"
:
-0.28
19824
2
,
"text"
:
"."
},
{
"id"
:
401
,
"logprob"
:
-7.9
0625
,
"logprob"
:
-7.9
179688
,
"text"
:
" G"
},
{
"id"
:
6013
,
"logprob"
:
-2.2
6562
5
,
"logprob"
:
-2.2
73437
5
,
"text"
:
"ira"
},
{
"id"
:
694
,
"logprob"
:
-0.6
40625
,
"logprob"
:
-0.6
230469
,
"text"
:
"ft"
},
{
"id"
:
1480
,
"logprob"
:
-0.20
3125
,
"logprob"
:
-0.20
947266
,
"text"
:
"ron"
},
{
"id"
:
9369
,
"logprob"
:
-4.5
312
5
,
"logprob"
:
-4.5
54687
5
,
"text"
:
" believes"
},
{
"id"
:
455
,
"logprob"
:
-4.5
6
25
,
"logprob"
:
-4.5
7031
25
,
"text"
:
" all"
},
{
"id"
:
599
,
"logprob"
:
-2.7
5
,
"logprob"
:
-2.7
382812
,
"text"
:
" other"
},
{
"id"
:
5632
,
"logprob"
:
-0.21
875
,
"logprob"
:
-0.21
948242
,
"text"
:
" animals"
},
{
"id"
:
362
,
"logprob"
:
-0.76
171875
,
"logprob"
:
-0.76
61133
,
"text"
:
" are"
},
{
"id"
:
23981
,
"logprob"
:
-4.9
6875
,
"logprob"
:
-4.9
960938
,
"text"
:
" irrelevant"
},
{
"id"
:
635
,
"logprob"
:
-4.2
18
75
,
"logprob"
:
-4.2
343
75
,
"text"
:
" when"
},
{
"id"
:
4354
,
"logprob"
:
-0.51
953125
,
"logprob"
:
-0.51
31836
,
"text"
:
" compared"
},
{
"id"
:
271
,
"logprob"
:
-0.1035
15625
,
"logprob"
:
-0.1035
7666
,
"text"
:
" to"
},
{
"id"
:
248
,
"logprob"
:
-0.58
984375
,
"logprob"
:
-0.58
447266
,
"text"
:
" the"
},
{
"id"
:
21735
,
"logprob"
:
-3.68
75
,
"logprob"
:
-3.68
16406
,
"text"
:
" glorious"
},
{
"id"
:
64398
,
"logprob"
:
-1.8
35937
5
,
"logprob"
:
-1.8
20312
5
,
"text"
:
" majesty"
},
{
"id"
:
275
,
"logprob"
:
-0.2
4316406
,
"logprob"
:
-0.2
3583984
,
"text"
:
" of"
},
{
"id"
:
248
,
"logprob"
:
-0.35
15625
,
"logprob"
:
-0.35
473633
,
"text"
:
" the"
},
{
"id"
:
26680
,
"logprob"
:
-0.24
414062
,
"logprob"
:
-0.24
572754
,
"text"
:
" gir"
},
{
"id"
:
23226
,
"logprob"
:
-0.0
3100586
,
"logprob"
:
-0.0
29586792
,
"text"
:
"affe"
},
{
"id"
:
25
,
"logprob"
:
-0.173
82812
,
"logprob"
:
-0.173
46191
,
"text"
:
"."
},
{
"id"
:
193
,
"logprob"
:
-1.39
84375
,
"logprob"
:
-1.39
45312
,
"text"
:
"
\n
"
},
{
...
...
@@ -625,67 +625,67 @@
},
{
"id"
:
23
,
"logprob"
:
-
1.0
,
"logprob"
:
-
0.99316406
,
"text"
:
","
},
{
"id"
:
29033
,
"logprob"
:
-2.2
1875
,
"logprob"
:
-2.2
324219
,
"text"
:
" Gir"
},
{
"id"
:
1622
,
"logprob"
:
-0.10
644531
,
"logprob"
:
-0.10
797119
,
"text"
:
"af"
},
{
"id"
:
249
,
"logprob"
:
-0.04
1992188
,
"logprob"
:
-0.04
248047
,
"text"
:
"at"
},
{
"id"
:
1480
,
"logprob"
:
-0.002
5024414
,
"logprob"
:
-0.002
4814606
,
"text"
:
"ron"
},
{
"id"
:
12
,
"logprob"
:
-1.42
96875
,
"logprob"
:
-1.42
77344
,
"text"
:
"!"
},
{
"id"
:
193
,
"logprob"
:
-1.10
15625
,
"logprob"
:
-1.10
05859
,
"text"
:
"
\n
"
},
{
"id"
:
50
,
"logprob"
:
-0.05
810547
,
"logprob"
:
-0.05
6884766
,
"text"
:
"G"
},
{
"id"
:
330
,
"logprob"
:
-0.1
2597656
,
"logprob"
:
-0.1
315918
,
"text"
:
"ir"
},
{
"id"
:
1622
,
"logprob"
:
-0.0070
80078
,
"logprob"
:
-0.0070
95337
,
"text"
:
"af"
},
{
"id"
:
249
,
"logprob"
:
-0.008
300781
,
"logprob"
:
-0.008
44574
,
"text"
:
"at"
},
{
"id"
:
1480
,
"logprob"
:
-0.00068
66455
,
"logprob"
:
-0.00068
998337
,
"text"
:
"ron"
},
{
"id"
:
37
,
"logprob"
:
-0.0074
157715
,
"logprob"
:
-0.0074
768066
,
"text"
:
":"
}
],
...
...
@@ -693,67 +693,67 @@
"tokens"
:
[
{
"id"
:
23090
,
"logprob"
:
-1.82
03125
,
"logprob"
:
-1.82
51953
,
"special"
:
false
,
"text"
:
" Hello"
},
{
"id"
:
23
,
"logprob"
:
-0.3
2226562
,
"logprob"
:
-0.3
1762695
,
"special"
:
false
,
"text"
:
","
},
{
"id"
:
8156
,
"logprob"
:
-0.238
28125
,
"logprob"
:
-0.238
8916
,
"special"
:
false
,
"text"
:
" Daniel"
},
{
"id"
:
12
,
"logprob"
:
-0.5
859375
,
"logprob"
:
-0.5
698242
,
"special"
:
false
,
"text"
:
"!"
},
{
"id"
:
193
,
"logprob"
:
-0.61
71875
,
"logprob"
:
-0.61
52344
,
"special"
:
false
,
"text"
:
"
\n
"
},
{
"id"
:
23626
,
"logprob"
:
-0.
39648438
,
"logprob"
:
-0.
42211914
,
"special"
:
false
,
"text"
:
"Daniel"
},
{
"id"
:
37
,
"logprob"
:
-0.00233
45947
,
"logprob"
:
-0.00233
6502
,
"special"
:
false
,
"text"
:
":"
},
{
"id"
:
295
,
"logprob"
:
-2.0
78125
,
"id"
:
1634
,
"logprob"
:
-2.0
605469
,
"special"
:
false
,
"text"
:
"
I
"
"text"
:
"
What
"
},
{
"id"
:
18
,
"logprob"
:
-1.
453125
,
"logprob"
:
-1.
5292969
,
"special"
:
false
,
"text"
:
"'"
},
{
"id"
:
88
,
"logprob"
:
-0.
47460938
,
"id"
:
94
,
"logprob"
:
-0.
007926941
,
"special"
:
false
,
"text"
:
"
m
"
"text"
:
"
s
"
}
]
},
"generated_text"
:
" Hello, Daniel!
\n
Daniel:
I'm
"
"generated_text"
:
" Hello, Daniel!
\n
Daniel:
What's
"
},
{
"details"
:
{
...
...
@@ -768,22 +768,22 @@
},
{
"id"
:
330
,
"logprob"
:
-5.
71
875
,
"logprob"
:
-5.
96
875
,
"text"
:
"ir"
},
{
"id"
:
1622
,
"logprob"
:
-5.6
2
5
,
"logprob"
:
-5.6
17187
5
,
"text"
:
"af"
},
{
"id"
:
249
,
"logprob"
:
-6.5
3125
,
"logprob"
:
-6.5
,
"text"
:
"at"
},
{
"id"
:
1480
,
"logprob"
:
-8.0
6
25
,
"logprob"
:
-8.0
7031
25
,
"text"
:
"ron"
},
{
...
...
@@ -793,197 +793,197 @@
},
{
"id"
:
23866
,
"logprob"
:
-9.
62
5
,
"logprob"
:
-9.
5937
5
,
"text"
:
" obsessed"
},
{
"id"
:
335
,
"logprob"
:
-0.048
58
3984
,
"logprob"
:
-0.048
3
3984
4
,
"text"
:
" with"
},
{
"id"
:
26680
,
"logprob"
:
-
3.984375
,
"logprob"
:
-
4.0
,
"text"
:
" gir"
},
{
"id"
:
1903
,
"logprob"
:
-0.07
6171875
,
"logprob"
:
-0.07
531738
,
"text"
:
"aff"
},
{
"id"
:
255
,
"logprob"
:
-0.006
683349
6
,
"logprob"
:
-0.006
79397
6
,
"text"
:
"es"
},
{
"id"
:
23
,
"logprob"
:
-1.54
6875
,
"logprob"
:
-1.54
78516
,
"text"
:
","
},
{
"id"
:
248
,
"logprob"
:
-4.3
4375
,
"logprob"
:
-4.3
320312
,
"text"
:
" the"
},
{
"id"
:
758
,
"logprob"
:
-3.73
4375
,
"logprob"
:
-3.73
63281
,
"text"
:
" most"
},
{
"id"
:
21735
,
"logprob"
:
-5.12
5
,
"logprob"
:
-5.1
13281
2
,
"text"
:
" glorious"
},
{
"id"
:
5985
,
"logprob"
:
-2.0
78125
,
"logprob"
:
-2.0
957031
,
"text"
:
" animal"
},
{
"id"
:
313
,
"logprob"
:
-1.1
953125
,
"logprob"
:
-1.1
835938
,
"text"
:
" on"
},
{
"id"
:
248
,
"logprob"
:
-0.7
8125
,
"logprob"
:
-0.7
7685547
,
"text"
:
" the"
},
{
"id"
:
1936
,
"logprob"
:
-2.3
90625
,
"logprob"
:
-2.3
808594
,
"text"
:
" face"
},
{
"id"
:
275
,
"logprob"
:
-0.0044
555664
,
"logprob"
:
-0.0044
36493
,
"text"
:
" of"
},
{
"id"
:
414
,
"logprob"
:
-1.9
84375
,
"logprob"
:
-1.9
638672
,
"text"
:
" this"
},
{
"id"
:
6490
,
"logprob"
:
-2.0
3125
,
"logprob"
:
-2.0
449219
,
"text"
:
" Earth"
},
{
"id"
:
25
,
"logprob"
:
-0.28
32031
2
,
"logprob"
:
-0.28
19824
2
,
"text"
:
"."
},
{
"id"
:
401
,
"logprob"
:
-7.9
0625
,
"logprob"
:
-7.9
179688
,
"text"
:
" G"
},
{
"id"
:
6013
,
"logprob"
:
-2.2
6562
5
,
"logprob"
:
-2.2
73437
5
,
"text"
:
"ira"
},
{
"id"
:
694
,
"logprob"
:
-0.6
40625
,
"logprob"
:
-0.6
230469
,
"text"
:
"ft"
},
{
"id"
:
1480
,
"logprob"
:
-0.20
3125
,
"logprob"
:
-0.20
947266
,
"text"
:
"ron"
},
{
"id"
:
9369
,
"logprob"
:
-4.5
312
5
,
"logprob"
:
-4.5
54687
5
,
"text"
:
" believes"
},
{
"id"
:
455
,
"logprob"
:
-4.5
6
25
,
"logprob"
:
-4.5
7031
25
,
"text"
:
" all"
},
{
"id"
:
599
,
"logprob"
:
-2.7
5
,
"logprob"
:
-2.7
382812
,
"text"
:
" other"
},
{
"id"
:
5632
,
"logprob"
:
-0.21
875
,
"logprob"
:
-0.21
948242
,
"text"
:
" animals"
},
{
"id"
:
362
,
"logprob"
:
-0.76
171875
,
"logprob"
:
-0.76
61133
,
"text"
:
" are"
},
{
"id"
:
23981
,
"logprob"
:
-4.9
6875
,
"logprob"
:
-4.9
960938
,
"text"
:
" irrelevant"
},
{
"id"
:
635
,
"logprob"
:
-4.2
18
75
,
"logprob"
:
-4.2
343
75
,
"text"
:
" when"
},
{
"id"
:
4354
,
"logprob"
:
-0.51
953125
,
"logprob"
:
-0.51
31836
,
"text"
:
" compared"
},
{
"id"
:
271
,
"logprob"
:
-0.1035
15625
,
"logprob"
:
-0.1035
7666
,
"text"
:
" to"
},
{
"id"
:
248
,
"logprob"
:
-0.58
984375
,
"logprob"
:
-0.58
447266
,
"text"
:
" the"
},
{
"id"
:
21735
,
"logprob"
:
-3.68
75
,
"logprob"
:
-3.68
16406
,
"text"
:
" glorious"
},
{
"id"
:
64398
,
"logprob"
:
-1.8
35937
5
,
"logprob"
:
-1.8
20312
5
,
"text"
:
" majesty"
},
{
"id"
:
275
,
"logprob"
:
-0.2
4316406
,
"logprob"
:
-0.2
3583984
,
"text"
:
" of"
},
{
"id"
:
248
,
"logprob"
:
-0.35
15625
,
"logprob"
:
-0.35
473633
,
"text"
:
" the"
},
{
"id"
:
26680
,
"logprob"
:
-0.24
414062
,
"logprob"
:
-0.24
572754
,
"text"
:
" gir"
},
{
"id"
:
23226
,
"logprob"
:
-0.0
3100586
,
"logprob"
:
-0.0
29586792
,
"text"
:
"affe"
},
{
"id"
:
25
,
"logprob"
:
-0.173
82812
,
"logprob"
:
-0.173
46191
,
"text"
:
"."
},
{
"id"
:
193
,
"logprob"
:
-1.39
84375
,
"logprob"
:
-1.39
45312
,
"text"
:
"
\n
"
},
{
...
...
@@ -1003,67 +1003,67 @@
},
{
"id"
:
23
,
"logprob"
:
-
1.0
,
"logprob"
:
-
0.99316406
,
"text"
:
","
},
{
"id"
:
29033
,
"logprob"
:
-2.2
1875
,
"logprob"
:
-2.2
324219
,
"text"
:
" Gir"
},
{
"id"
:
1622
,
"logprob"
:
-0.10
644531
,
"logprob"
:
-0.10
797119
,
"text"
:
"af"
},
{
"id"
:
249
,
"logprob"
:
-0.04
1992188
,
"logprob"
:
-0.04
248047
,
"text"
:
"at"
},
{
"id"
:
1480
,
"logprob"
:
-0.002
5024414
,
"logprob"
:
-0.002
4814606
,
"text"
:
"ron"
},
{
"id"
:
12
,
"logprob"
:
-1.42
96875
,
"logprob"
:
-1.42
77344
,
"text"
:
"!"
},
{
"id"
:
193
,
"logprob"
:
-1.10
15625
,
"logprob"
:
-1.10
05859
,
"text"
:
"
\n
"
},
{
"id"
:
50
,
"logprob"
:
-0.05
810547
,
"logprob"
:
-0.05
6884766
,
"text"
:
"G"
},
{
"id"
:
330
,
"logprob"
:
-0.1
2597656
,
"logprob"
:
-0.1
315918
,
"text"
:
"ir"
},
{
"id"
:
1622
,
"logprob"
:
-0.0070
80078
,
"logprob"
:
-0.0070
95337
,
"text"
:
"af"
},
{
"id"
:
249
,
"logprob"
:
-0.008
300781
,
"logprob"
:
-0.008
44574
,
"text"
:
"at"
},
{
"id"
:
1480
,
"logprob"
:
-0.00068
66455
,
"logprob"
:
-0.00068
998337
,
"text"
:
"ron"
},
{
"id"
:
37
,
"logprob"
:
-0.0074
157715
,
"logprob"
:
-0.0074
768066
,
"text"
:
":"
}
],
...
...
@@ -1071,67 +1071,67 @@
"tokens"
:
[
{
"id"
:
23090
,
"logprob"
:
-1.82
03125
,
"logprob"
:
-1.82
51953
,
"special"
:
false
,
"text"
:
" Hello"
},
{
"id"
:
23
,
"logprob"
:
-0.3
2226562
,
"logprob"
:
-0.3
1762695
,
"special"
:
false
,
"text"
:
","
},
{
"id"
:
8156
,
"logprob"
:
-0.238
28125
,
"logprob"
:
-0.238
8916
,
"special"
:
false
,
"text"
:
" Daniel"
},
{
"id"
:
12
,
"logprob"
:
-0.5
859375
,
"logprob"
:
-0.5
698242
,
"special"
:
false
,
"text"
:
"!"
},
{
"id"
:
193
,
"logprob"
:
-0.61
71875
,
"logprob"
:
-0.61
52344
,
"special"
:
false
,
"text"
:
"
\n
"
},
{
"id"
:
23626
,
"logprob"
:
-0.
39648438
,
"logprob"
:
-0.
42211914
,
"special"
:
false
,
"text"
:
"Daniel"
},
{
"id"
:
37
,
"logprob"
:
-0.00233
45947
,
"logprob"
:
-0.00233
6502
,
"special"
:
false
,
"text"
:
":"
},
{
"id"
:
295
,
"logprob"
:
-2.0
78125
,
"id"
:
1634
,
"logprob"
:
-2.0
605469
,
"special"
:
false
,
"text"
:
"
I
"
"text"
:
"
What
"
},
{
"id"
:
18
,
"logprob"
:
-1.
453125
,
"logprob"
:
-1.
5292969
,
"special"
:
false
,
"text"
:
"'"
},
{
"id"
:
88
,
"logprob"
:
-0.
47460938
,
"id"
:
94
,
"logprob"
:
-0.
007926941
,
"special"
:
false
,
"text"
:
"
m
"
"text"
:
"
s
"
}
]
},
"generated_text"
:
" Hello, Daniel!
\n
Daniel:
I'm
"
"generated_text"
:
" Hello, Daniel!
\n
Daniel:
What's
"
},
{
"details"
:
{
...
...
@@ -1146,22 +1146,22 @@
},
{
"id"
:
330
,
"logprob"
:
-5.
71
875
,
"logprob"
:
-5.
96
875
,
"text"
:
"ir"
},
{
"id"
:
1622
,
"logprob"
:
-5.6
2
5
,
"logprob"
:
-5.6
17187
5
,
"text"
:
"af"
},
{
"id"
:
249
,
"logprob"
:
-6.5
3125
,
"logprob"
:
-6.5
,
"text"
:
"at"
},
{
"id"
:
1480
,
"logprob"
:
-8.0
6
25
,
"logprob"
:
-8.0
7031
25
,
"text"
:
"ron"
},
{
...
...
@@ -1171,197 +1171,197 @@
},
{
"id"
:
23866
,
"logprob"
:
-9.
62
5
,
"logprob"
:
-9.
5937
5
,
"text"
:
" obsessed"
},
{
"id"
:
335
,
"logprob"
:
-0.048
58
3984
,
"logprob"
:
-0.048
3
3984
4
,
"text"
:
" with"
},
{
"id"
:
26680
,
"logprob"
:
-
3.984375
,
"logprob"
:
-
4.0
,
"text"
:
" gir"
},
{
"id"
:
1903
,
"logprob"
:
-0.07
6171875
,
"logprob"
:
-0.07
531738
,
"text"
:
"aff"
},
{
"id"
:
255
,
"logprob"
:
-0.006
683349
6
,
"logprob"
:
-0.006
79397
6
,
"text"
:
"es"
},
{
"id"
:
23
,
"logprob"
:
-1.54
6875
,
"logprob"
:
-1.54
78516
,
"text"
:
","
},
{
"id"
:
248
,
"logprob"
:
-4.3
4375
,
"logprob"
:
-4.3
320312
,
"text"
:
" the"
},
{
"id"
:
758
,
"logprob"
:
-3.73
4375
,
"logprob"
:
-3.73
63281
,
"text"
:
" most"
},
{
"id"
:
21735
,
"logprob"
:
-5.12
5
,
"logprob"
:
-5.1
13281
2
,
"text"
:
" glorious"
},
{
"id"
:
5985
,
"logprob"
:
-2.0
78125
,
"logprob"
:
-2.0
957031
,
"text"
:
" animal"
},
{
"id"
:
313
,
"logprob"
:
-1.1
953125
,
"logprob"
:
-1.1
835938
,
"text"
:
" on"
},
{
"id"
:
248
,
"logprob"
:
-0.7
8125
,
"logprob"
:
-0.7
7685547
,
"text"
:
" the"
},
{
"id"
:
1936
,
"logprob"
:
-2.3
90625
,
"logprob"
:
-2.3
808594
,
"text"
:
" face"
},
{
"id"
:
275
,
"logprob"
:
-0.0044
555664
,
"logprob"
:
-0.0044
36493
,
"text"
:
" of"
},
{
"id"
:
414
,
"logprob"
:
-1.9
84375
,
"logprob"
:
-1.9
638672
,
"text"
:
" this"
},
{
"id"
:
6490
,
"logprob"
:
-2.0
3125
,
"logprob"
:
-2.0
449219
,
"text"
:
" Earth"
},
{
"id"
:
25
,
"logprob"
:
-0.28
32031
2
,
"logprob"
:
-0.28
19824
2
,
"text"
:
"."
},
{
"id"
:
401
,
"logprob"
:
-7.9
0625
,
"logprob"
:
-7.9
179688
,
"text"
:
" G"
},
{
"id"
:
6013
,
"logprob"
:
-2.2
6562
5
,
"logprob"
:
-2.2
73437
5
,
"text"
:
"ira"
},
{
"id"
:
694
,
"logprob"
:
-0.6
40625
,
"logprob"
:
-0.6
230469
,
"text"
:
"ft"
},
{
"id"
:
1480
,
"logprob"
:
-0.20
3125
,
"logprob"
:
-0.20
947266
,
"text"
:
"ron"
},
{
"id"
:
9369
,
"logprob"
:
-4.5
312
5
,
"logprob"
:
-4.5
54687
5
,
"text"
:
" believes"
},
{
"id"
:
455
,
"logprob"
:
-4.5
6
25
,
"logprob"
:
-4.5
7031
25
,
"text"
:
" all"
},
{
"id"
:
599
,
"logprob"
:
-2.7
5
,
"logprob"
:
-2.7
382812
,
"text"
:
" other"
},
{
"id"
:
5632
,
"logprob"
:
-0.21
875
,
"logprob"
:
-0.21
948242
,
"text"
:
" animals"
},
{
"id"
:
362
,
"logprob"
:
-0.76
171875
,
"logprob"
:
-0.76
61133
,
"text"
:
" are"
},
{
"id"
:
23981
,
"logprob"
:
-4.9
6875
,
"logprob"
:
-4.9
960938
,
"text"
:
" irrelevant"
},
{
"id"
:
635
,
"logprob"
:
-4.2
18
75
,
"logprob"
:
-4.2
343
75
,
"text"
:
" when"
},
{
"id"
:
4354
,
"logprob"
:
-0.51
953125
,
"logprob"
:
-0.51
31836
,
"text"
:
" compared"
},
{
"id"
:
271
,
"logprob"
:
-0.1035
15625
,
"logprob"
:
-0.1035
7666
,
"text"
:
" to"
},
{
"id"
:
248
,
"logprob"
:
-0.58
984375
,
"logprob"
:
-0.58
447266
,
"text"
:
" the"
},
{
"id"
:
21735
,
"logprob"
:
-3.68
75
,
"logprob"
:
-3.68
16406
,
"text"
:
" glorious"
},
{
"id"
:
64398
,
"logprob"
:
-1.8
35937
5
,
"logprob"
:
-1.8
20312
5
,
"text"
:
" majesty"
},
{
"id"
:
275
,
"logprob"
:
-0.2
4316406
,
"logprob"
:
-0.2
3583984
,
"text"
:
" of"
},
{
"id"
:
248
,
"logprob"
:
-0.35
15625
,
"logprob"
:
-0.35
473633
,
"text"
:
" the"
},
{
"id"
:
26680
,
"logprob"
:
-0.24
414062
,
"logprob"
:
-0.24
572754
,
"text"
:
" gir"
},
{
"id"
:
23226
,
"logprob"
:
-0.0
3100586
,
"logprob"
:
-0.0
29586792
,
"text"
:
"affe"
},
{
"id"
:
25
,
"logprob"
:
-0.173
82812
,
"logprob"
:
-0.173
46191
,
"text"
:
"."
},
{
"id"
:
193
,
"logprob"
:
-1.39
84375
,
"logprob"
:
-1.39
45312
,
"text"
:
"
\n
"
},
{
...
...
@@ -1381,67 +1381,67 @@
},
{
"id"
:
23
,
"logprob"
:
-
1.0
,
"logprob"
:
-
0.99316406
,
"text"
:
","
},
{
"id"
:
29033
,
"logprob"
:
-2.2
1875
,
"logprob"
:
-2.2
324219
,
"text"
:
" Gir"
},
{
"id"
:
1622
,
"logprob"
:
-0.10
644531
,
"logprob"
:
-0.10
797119
,
"text"
:
"af"
},
{
"id"
:
249
,
"logprob"
:
-0.04
1992188
,
"logprob"
:
-0.04
248047
,
"text"
:
"at"
},
{
"id"
:
1480
,
"logprob"
:
-0.002
5024414
,
"logprob"
:
-0.002
4814606
,
"text"
:
"ron"
},
{
"id"
:
12
,
"logprob"
:
-1.42
96875
,
"logprob"
:
-1.42
77344
,
"text"
:
"!"
},
{
"id"
:
193
,
"logprob"
:
-1.10
15625
,
"logprob"
:
-1.10
05859
,
"text"
:
"
\n
"
},
{
"id"
:
50
,
"logprob"
:
-0.05
810547
,
"logprob"
:
-0.05
6884766
,
"text"
:
"G"
},
{
"id"
:
330
,
"logprob"
:
-0.1
2597656
,
"logprob"
:
-0.1
315918
,
"text"
:
"ir"
},
{
"id"
:
1622
,
"logprob"
:
-0.0070
80078
,
"logprob"
:
-0.0070
95337
,
"text"
:
"af"
},
{
"id"
:
249
,
"logprob"
:
-0.008
300781
,
"logprob"
:
-0.008
44574
,
"text"
:
"at"
},
{
"id"
:
1480
,
"logprob"
:
-0.00068
66455
,
"logprob"
:
-0.00068
998337
,
"text"
:
"ron"
},
{
"id"
:
37
,
"logprob"
:
-0.0074
157715
,
"logprob"
:
-0.0074
768066
,
"text"
:
":"
}
],
...
...
@@ -1449,66 +1449,66 @@
"tokens"
:
[
{
"id"
:
23090
,
"logprob"
:
-1.82
03125
,
"logprob"
:
-1.82
51953
,
"special"
:
false
,
"text"
:
" Hello"
},
{
"id"
:
23
,
"logprob"
:
-0.3
2226562
,
"logprob"
:
-0.3
1762695
,
"special"
:
false
,
"text"
:
","
},
{
"id"
:
8156
,
"logprob"
:
-0.238
28125
,
"logprob"
:
-0.238
8916
,
"special"
:
false
,
"text"
:
" Daniel"
},
{
"id"
:
12
,
"logprob"
:
-0.5
859375
,
"logprob"
:
-0.5
698242
,
"special"
:
false
,
"text"
:
"!"
},
{
"id"
:
193
,
"logprob"
:
-0.61
71875
,
"logprob"
:
-0.61
52344
,
"special"
:
false
,
"text"
:
"
\n
"
},
{
"id"
:
23626
,
"logprob"
:
-0.
39648438
,
"logprob"
:
-0.
42211914
,
"special"
:
false
,
"text"
:
"Daniel"
},
{
"id"
:
37
,
"logprob"
:
-0.00233
45947
,
"logprob"
:
-0.00233
6502
,
"special"
:
false
,
"text"
:
":"
},
{
"id"
:
295
,
"logprob"
:
-2.0
78125
,
"id"
:
1634
,
"logprob"
:
-2.0
605469
,
"special"
:
false
,
"text"
:
"
I
"
"text"
:
"
What
"
},
{
"id"
:
18
,
"logprob"
:
-1.
453125
,
"logprob"
:
-1.
5292969
,
"special"
:
false
,
"text"
:
"'"
},
{
"id"
:
88
,
"logprob"
:
-0.
47460938
,
"id"
:
94
,
"logprob"
:
-0.
007926941
,
"special"
:
false
,
"text"
:
"
m
"
"text"
:
"
s
"
}
]
},
"generated_text"
:
" Hello, Daniel!
\n
Daniel:
I'm
"
"generated_text"
:
" Hello, Daniel!
\n
Daniel:
What's
"
}
]
server/text_generation_server/models/flash_rw.py
View file @
87dc034b
...
...
@@ -37,7 +37,7 @@ class FlashRW(FlashCausalLM):
):
if
torch
.
cuda
.
is_available
():
device
=
torch
.
device
(
"cuda"
)
dtype
=
torch
.
b
float16
dtype
=
torch
.
float16
else
:
raise
NotImplementedError
(
"RW is only available on GPU"
)
...
...
@@ -124,7 +124,7 @@ class FlashRWSharded(FlashRW):
self
.
process_group
,
rank
,
world_size
=
initialize_torch_distributed
()
if
torch
.
cuda
.
is_available
():
device
=
torch
.
device
(
f
"cuda:
{
rank
}
"
)
dtype
=
torch
.
b
float16
dtype
=
torch
.
float16
else
:
raise
NotImplementedError
(
"FlashRW is only available on GPU"
)
...
...
server/text_generation_server/models/rw.py
View file @
87dc034b
...
...
@@ -16,7 +16,7 @@ class RW(CausalLM):
):
if
torch
.
cuda
.
is_available
():
device
=
torch
.
device
(
"cuda"
)
dtype
=
torch
.
b
float16
dtype
=
torch
.
float16
else
:
if
quantize
:
raise
ValueError
(
"quantization is not available on CPU"
)
...
...
server/text_generation_server/utils/hub.py
View file @
87dc034b
...
...
@@ -23,7 +23,11 @@ def weight_hub_files(
"""Get the weights filenames on the hub"""
api
=
HfApi
()
info
=
api
.
model_info
(
model_id
,
revision
=
revision
)
filenames
=
[
s
.
rfilename
for
s
in
info
.
siblings
if
s
.
rfilename
.
endswith
(
extension
)]
filenames
=
[
s
.
rfilename
for
s
in
info
.
siblings
if
s
.
rfilename
.
endswith
(
extension
)
and
len
(
s
.
rfilename
.
split
(
"/"
))
==
1
]
if
not
filenames
:
raise
EntryNotFoundError
(
...
...
@@ -130,24 +134,31 @@ def download_weights(
)
->
List
[
Path
]:
"""Download the safetensors files from the hub"""
def
download_file
(
filename
):
def
download_file
(
filename
,
tries
=
5
):
local_file
=
try_to_load_from_cache
(
model_id
,
revision
,
filename
)
if
local_file
is
not
None
:
logger
.
info
(
f
"File
{
filename
}
already present in cache."
)
return
Path
(
local_file
)
logger
.
info
(
f
"Download file:
{
filename
}
"
)
start_time
=
time
.
time
()
local_file
=
hf_hub_download
(
filename
=
filename
,
repo_id
=
model_id
,
revision
=
revision
,
local_files_only
=
False
,
)
logger
.
info
(
f
"Downloaded
{
local_file
}
in
{
timedelta
(
seconds
=
int
(
time
.
time
()
-
start_time
))
}
."
)
return
Path
(
local_file
)
for
i
in
range
(
tries
):
try
:
logger
.
info
(
f
"Download file:
{
filename
}
"
)
start_time
=
time
.
time
()
local_file
=
hf_hub_download
(
filename
=
filename
,
repo_id
=
model_id
,
revision
=
revision
,
local_files_only
=
False
,
)
logger
.
info
(
f
"Downloaded
{
local_file
}
in
{
timedelta
(
seconds
=
int
(
time
.
time
()
-
start_time
))
}
."
)
return
Path
(
local_file
)
except
Exception
as
e
:
if
i
+
1
==
tries
:
raise
e
logger
.
error
(
e
)
logger
.
info
(
f
"Retry
{
i
+
1
}
/
{
tries
-
1
}
"
)
# We do this instead of using tqdm because we want to parse the logs with the launcher
start_time
=
time
.
time
()
...
...
Write
Preview
Markdown
is supported
0%
Try again
or
attach a new file
.
Attach a file
Cancel
You are about to add
0
people
to the discussion. Proceed with caution.
Finish editing this message first!
Cancel
Please
register
or
sign in
to comment