huggingface/text-generation-inference

Pull Requests Commits

Add tgi_batch_current_size and tgi_batch_current_size as response header

co42 committed 364 days ago

b4187d60

Add fp8 kv cache for ROCm (#2856)

mht-sharma committed 364 days ago

Verified c20025db

Moving to `uv` instead of `poetry`. (#2919)

Narsil committed 364 days ago

Verified de19e7e8

nix: update to PyTorch 2.5.1 (#2921)

danieldk committed 364 days ago

Verified d61f14f2

Flash decoding kernel adding and prefill-chunking and prefix caching enabling in intel cpu/xpu (#2815)

sywangyi committed 364 days ago

Verified 88514416

feat: improve star coder to support multi lora layers (#2883)

drbh committed 1 year ago

Verified 82f6ea1b

Do not convert weight scale to e4m3fnuz on CUDA (#2917)

danieldk committed 1 year ago

Verified 5f78ec32

Upgrading bitsandbytes. (#2910)

Narsil committed 1 year ago

Verified 922cc38f

Removing the github runner. (#2912)

Narsil committed 1 year ago

Verified 120bd3e3

Fix typo in TPU docs (#2911)

baptistecolle committed 1 year ago

Verified 1470aec9

Upgrading our rustc version. (#2908)

Narsil committed 1 year ago

Verified 203cade2

:memo: add guide on using TPU with TGI in the docs (#2907)

baptistecolle committed 1 year ago

Verified 46994b34

Fix `docker run` in `README.md` (#2861)

alvarobartt committed 1 year ago

Verified dc9b8e98

docs(conceptual/speculation): available links Train Medusa (#2863)

guspan-tanadi committed 1 year ago

Verified 3c7ae48f

Baichuan2-13B does not have max_position_embeddings in config (#2903)

sywangyi committed 1 year ago

Verified cc8b9650

Enable FP8 Per-Tensor Scales and Integrate Marlin/MoE Kernels Repo for ROCm (#2825)

mht-sharma committed 1 year ago

Verified e07acc7f

Add Flash decoding kernel ROCm (#2855)

mht-sharma committed 1 year ago

Verified 880ab9c2

fix crash in torch2.6 if TP=1 (#2885)

sywangyi committed 1 year ago

Verified 1660154a

Update using_guidance.md (#2901)

nbroad1881 committed 1 year ago

Verified 2e22164d

Add possible variants for A100 and H100 GPUs for auto-detecting flops (#2837)

lazariv committed 1 year ago

Verified 83624a07

chore: Update jsonschema to 0.28.0 (#2870)

Stranger6667 committed 1 year ago

Verified 01067f8b

Update to marlin-kernels 0.3.7 (#2882)

danieldk committed 1 year ago

Verified 4f7e00f4

Improve vlm support (add idefics3 support) (#2437)

drbh committed 1 year ago

Verified da5ab467

Basic flashinfer 0.2 support (#2862)

danieldk committed 1 year ago

Verified a9c7d2e3

update ipex xpu to fix issue in ARC770 (#2884)

sywangyi committed 1 year ago

Verified afb6c728

chore: fixed some typos and attribute issues in README (#2891)

ruidazeng committed 1 year ago

Verified d37a43e5

fix: include add_special_tokens in kserve request (#2859)

drbh committed 1 year ago

Verified 23bc38b1

change xpu lib download link (#2852)

sywangyi committed 1 year ago

Verified ab5f6169

Update vllm kernels for ROCM (#2826)

mht-sharma committed 1 year ago

Verified 8f66d323

Qwen2-VL runtime error fix when prompted with multiple images (#2840)

janne-alatalo committed 1 year ago

Verified 7eeefa3b

Older