Chunyuan WU a5317b2fd3 [CPU] add optimizations for INT8 and FP8 DeepSeek (#6769) ...

Co-authored-by: Zheng, Beilei <beilei.zheng@intel.com>

2025-06-27 19:04:29 -07:00

..

configs

Hybrid kv cache for LLaMA4 (#6563)

2025-06-27 18:58:55 -07:00

connector

feat(remote_model): support variable remote backend for model loader (#3964)

2025-03-14 00:40:44 -07:00

constrained

Correctly abort the failed grammar requests & Improve the handling of abort (#6803)

2025-06-01 19:00:07 -07:00

disaggregation

Hybrid kv cache for LLaMA4 (#6563)

2025-06-27 18:58:55 -07:00

distributed

Fix All-Gather under world size one (#7219)

2025-06-20 14:57:34 -07:00

entrypoints

minor: 'role' must be system/assistant/tool, but case insensitive for now (#7499)

2025-06-25 02:11:03 -07:00

eplb_simulator

Minor add utility to read expert distribution recorder output (#7134)

2025-06-12 16:50:05 -07:00

function_call

feat(func_call): Add more check in BaseFormatDetector.parse_streaming_increment (#7479)

2025-06-23 11:15:47 -07:00

layers

[CPU] add optimizations for INT8 and FP8 DeepSeek (#6769)

2025-06-27 19:04:29 -07:00

lora

Refactor LoRAManager and LoRAMemoryPool state management logic for dynamic LoRA loading support (#7412)

2025-06-21 16:09:19 -07:00

managers

Hybrid kv cache for LLaMA4 (#6563)

2025-06-27 18:58:55 -07:00

mem_cache

Hybrid kv cache for LLaMA4 (#6563)

2025-06-27 18:58:55 -07:00

metrics

Correctly abort the failed grammar requests & Improve the handling of abort (#6803)

2025-06-01 19:00:07 -07:00

model_executor

Hybrid kv cache for LLaMA4 (#6563)

2025-06-27 18:58:55 -07:00

model_loader

Support multi-thread model weight loading (#7277)

2025-06-24 10:39:10 -07:00

models

[CPU] add optimizations for INT8 and FP8 DeepSeek (#6769)

2025-06-27 19:04:29 -07:00

multimodal

Move multimodal processors into a separate folder (#7581)

2025-06-27 11:58:24 -07:00

sampling

[Feature] Add Logit Bias (#6579)

2025-06-10 15:39:25 -07:00

speculative

Perormance: Enable cuda graph for dp idle batch (#7269)

2025-06-23 17:34:13 -07:00

_custom_ops.py

support qwen2 running on ascend npu device (#7022)

2025-06-17 11:24:10 -07:00

aio_rwlock.py

A better aio rwlock that guarantees the order (#2547)

2024-12-22 15:44:32 -08:00

code_completion_parser.py

feat(oai refactor): Replace openai_api with entrypoints/openai (#7351)

2025-06-21 13:21:06 -07:00

constants.py

Multi-Stage Awake: Support Resume and Pause KV Cache and Weights separately (#7099)

2025-06-19 00:56:37 -07:00

conversation.py

chore: remove vlm unnecessary import (#7541)

2025-06-26 01:38:15 -07:00

custom_op.py

npu fused op (#7386)

2025-06-25 01:54:20 -07:00

debug_utils.py

Add simple utility to dump tensors for debugging (#6815)

2025-06-02 11:15:31 -07:00

hf_transformers_utils.py

[New Model] Devstral support (#6547)

2025-05-26 19:27:48 -07:00

jinja_template_utils.py

feat(oai refactor): Replace openai_api with entrypoints/openai (#7351)

2025-06-21 13:21:06 -07:00

model_parallel.py

Make torch TP composable with torchao (#2436)

2024-12-11 04:21:42 -08:00

operations_strategy.py

feat: support compatibility between MTP and two-batch-overlap (#7225)

2025-06-27 01:10:27 -07:00

operations.py

Support overlapping two batches (#4068)

2025-05-24 17:39:07 -07:00

patch_torch.py

Fix torch.compile cacheing (#5259)

2025-04-10 18:08:45 -07:00

reasoning_parser.py

feat(oai refactor): Replace openai_api with entrypoints/openai (#7351)

2025-06-21 13:21:06 -07:00

server_args.py

Hybrid kv cache for LLaMA4 (#6563)

2025-06-27 18:58:55 -07:00

torch_memory_saver_adapter.py

Multi-Stage Awake: Support Resume and Pause KV Cache and Weights separately (#7099)

2025-06-19 00:56:37 -07:00

two_batch_overlap.py

feat: support compatibility between MTP and two-batch-overlap (#7225)

2025-06-27 01:10:27 -07:00

utils.py

Move multimodal processors into a separate folder (#7581)

2025-06-27 11:58:24 -07:00

warmup.py

Support penalty in overlap mode; return logprob with chunked prefill; improve benchmark scripts (#3988)

2025-03-03 00:12:04 -08:00