- 07 Jul, 2024 1 commit
-
-
Robert Shaw authored
Co-authored-by:Robert Shaw <rshaw@neuralmagic>
-
- 03 Jul, 2024 3 commits
-
-
Robert Shaw authored
-
Michael Goin authored
-
youkaichao authored
-
- 02 Jul, 2024 2 commits
-
-
Qubitium-ModelCloud authored
Co-authored-by:
Robert Shaw <rshaw@neuralmagic.com> Co-authored-by:
ZX <zx@lbx.dev>
-
Robert Shaw authored
Co-authored-by:
Michael Goin <michael@neuralmagic.com>
-
- 01 Jul, 2024 1 commit
-
-
youkaichao authored
-
- 30 Jun, 2024 2 commits
-
-
Robert Shaw authored
Co-authored-by: -
Dipika Sikka authored
-
- 29 Jun, 2024 1 commit
-
-
Robert Shaw authored
Co-authored-by:
-
- 28 Jun, 2024 1 commit
-
-
Robert Shaw authored
Co-authored-by:
-
- 25 Jun, 2024 1 commit
-
-
Dipika Sikka authored
-
- 20 Jun, 2024 1 commit
-
-
Tyler Michael Smith authored
-
- 19 Jun, 2024 1 commit
-
-
Dipika Sikka authored
[Misc] Add per channel support for static activation quantization; update w8a8 schemes to share base classes (#5650)
-
- 18 Jun, 2024 1 commit
-
-
Dipika Sikka authored
[Misc] Add channel-wise quantization support for w8a8 dynamic per token activation quantization (#5542)
-
- 17 Jun, 2024 1 commit
-
-
Dipika Sikka authored
-
- 15 Jun, 2024 1 commit
-
-
Cyrus Leung authored
-
- 14 Jun, 2024 2 commits
-
-
Robert Shaw authored
Co-authored-by:
Dipika Sikka <dipikasikka1@gmail.com>
-
Tyler Michael Smith authored
-
- 13 Jun, 2024 3 commits
-
-
Tyler Michael Smith authored
-
Tyler Michael Smith authored
Co-authored-by:
youkaichao <youkaichao@gmail.com> Co-authored-by:
zifeitong <zifei.tong@parasail.io> Co-authored-by:
Robert Shaw <114415538+robertgshaw2-neuralmagic@users.noreply.github.com>
-
Dipika Sikka authored
Co-authored-by:
-
- 10 Jun, 2024 1 commit
-
-
Dipika Sikka authored
Co-authored-by:
-
- 08 Jun, 2024 2 commits
-
-
Michael Goin authored
-
Cheng Li authored
Bug description: With torch 2.4.0.dev20240603+cu121, cutlass_fp8_supported outputs False, and the (capability, version) before the comparison is (90, 11111111112) This PR fixes the support check for FP8 CUTLASS ( cutlass_fp8_supported) which was introduced in https://github.com/vllm-project/vllm/pull/5183.
-
- 07 Jun, 2024 2 commits
-
-
Dipika Sikka authored
Co-authored-by:
Varun Sundar Rabindranath <varunsundar08@gmail.com> Co-authored-by:
Varun Sundar Rabindranath <varun@neuralmagic.com>
-
Tyler Michael Smith authored
Switching from torch._scaled_mm to vLLM's cutlass fp8 kernels when supported as we are seeing 5-15% improvement in e2e performance on neuralmagic/Meta-Llama-3-8B-Instruct-FP8 see https://docs.google.com/spreadsheets/d/1GiAnmzyGHgZ6zL_LDSTm35Bdrt4A8AaFEurDlISYYA4/ for some quick e2e benchmarks and #5144 for comparisons across different GEMM sizes.
-
- 05 Jun, 2024 1 commit
-
-
Cody Yu authored
-
- 03 Jun, 2024 1 commit
-
-
Tyler Michael Smith authored
-
- 01 Jun, 2024 2 commits
-
-
chenqianfzh authored
-
Tyler Michael Smith authored
-
- 31 May, 2024 1 commit
-
-
Robert Shaw authored
-
- 30 May, 2024 1 commit
-
-
Alexander Matveev authored
-
- 27 May, 2024 1 commit
-
-
sasha0552 authored
-
- 23 May, 2024 2 commits
-
-
Dipika Sikka authored
Co-authored-by:
-
Alexander Matveev authored
-
- 22 May, 2024 1 commit
-
-
Cody Yu authored
The 2nd PR for #4532. This PR supports loading FP8 kv-cache scaling factors from a FP8 checkpoint (with .kv_scale parameter).
-
- 19 May, 2024 1 commit
-
-
Alexander Matveev authored
-
- 16 May, 2024 2 commits
-
-
Alexander Matveev authored
Co-authored-by: -
Jinzhen Lin authored
-
