Commit 6d7187b

committed

[V1] Support ngram spec decode

Signed-off-by: ponix-j <657511300@qq.com>

1 parent 00e0243 commit 6d7187bCopy full SHA for 6d7187b

11 files changed

+1746

-11

lines changed

requirements-dev.txt
tests
- sample
  - __init__.py
  - test_rejection_sampler.py
- singlecard/spec_decode/e2e
  - test_spec_decode.py
vllm_ascend
- attention
  - attention_v1.py
- patch
  - __init__.py
  - worker/patch_common
    - __init__.py
    - patch_eagle.py
- sample
  - __init__.py
  - rejection_sampler.py
- worker
  - model_runner_v1.py

11 files changed

+1746

-11

lines changed

`‎requirements-dev.txt`

-Original file line number
+Diff line change
 types-jsonschema
 xgrammar
 zmq
 +numba

`‎tests/sample/init.py`

Whitespace-only changes.

Comments

(0)