| .. |
|
__init__.py
|
a5bfc2bc3d
VLM: add support for LLaVA-Onevision model (#1100)
|
hai 1 mes |
|
arctic.py
|
0dfa6b60ec
core: support logprobs with multi-step scheduling (#963)
|
hai 2 meses |
|
baichuan.py
|
0dfa6b60ec
core: support logprobs with multi-step scheduling (#963)
|
hai 2 meses |
|
bart.py
|
a985143768
core: add cuda graph support for encoder-decoder models (#1051)
|
hai 2 meses |
|
blip.py
|
7ce3174039
VLM: refactor blip models to support composite weight loading (#1098)
|
hai 1 mes |
|
blip2.py
|
7ce3174039
VLM: refactor blip models to support composite weight loading (#1098)
|
hai 1 mes |
|
bloom.py
|
0dfa6b60ec
core: support logprobs with multi-step scheduling (#963)
|
hai 2 meses |
|
chameleon.py
|
651678d2df
VLM: use `SequenceData.from_token_counts` to create dummy data (#1093)
|
hai 1 mes |
|
chatglm.py
|
0dfa6b60ec
core: support logprobs with multi-step scheduling (#963)
|
hai 2 meses |
|
clip.py
|
80be38ca6f
chore: expose phi3_v num_crops as an mm_processor_kwargs (#1117)
|
hai 1 mes |
|
commandr.py
|
135dfd648b
fix: LoRA support for Cohere and Jamba models (#1004)
|
hai 2 meses |
|
dbrx.py
|
b65449b5ad
moe: refactor DBRX experts to support FusedMoE (#1095)
|
hai 1 mes |
|
decilm.py
|
9022c6d869
remove progress_bar imports
|
hai 4 meses |
|
deepseek.py
|
0dfa6b60ec
core: support logprobs with multi-step scheduling (#963)
|
hai 2 meses |
|
deepseek_v2.py
|
0dfa6b60ec
core: support logprobs with multi-step scheduling (#963)
|
hai 2 meses |
|
eagle.py
|
0dfa6b60ec
core: support logprobs with multi-step scheduling (#963)
|
hai 2 meses |
|
exaone.py
|
0dfa6b60ec
core: support logprobs with multi-step scheduling (#963)
|
hai 2 meses |
|
falcon.py
|
0dfa6b60ec
core: support logprobs with multi-step scheduling (#963)
|
hai 2 meses |
|
fuyu.py
|
0dfa6b60ec
core: support logprobs with multi-step scheduling (#963)
|
hai 2 meses |
|
gemma.py
|
0dfa6b60ec
core: support logprobs with multi-step scheduling (#963)
|
hai 2 meses |
|
gemma2.py
|
b33cf04386
quants: add bitsandbytes support for gemma2 model (#1026)
|
hai 2 meses |
|
gpt2.py
|
0dfa6b60ec
core: support logprobs with multi-step scheduling (#963)
|
hai 2 meses |
|
gpt_bigcode.py
|
0dfa6b60ec
core: support logprobs with multi-step scheduling (#963)
|
hai 2 meses |
|
gpt_j.py
|
0dfa6b60ec
core: support logprobs with multi-step scheduling (#963)
|
hai 2 meses |
|
gpt_neox.py
|
0dfa6b60ec
core: support logprobs with multi-step scheduling (#963)
|
hai 2 meses |
|
granite.py
|
11f49b5341
fix: granite logit scale in logit computation (#1054)
|
hai 2 meses |
|
idefics2_vision_model.py
|
f1d0b77c92
[0.6.0] Release Candidate (#481)
|
hai 6 meses |
|
interfaces.py
|
0b8b407b6d
feat: support profiling with multiple multi-modal inputs per prompt (#712)
|
hai 6 meses |
|
intern_vit.py
|
f56d6b396a
vlm: fallback to SDPA for ViT models on CPU backend (#982)
|
hai 2 meses |
|
internlm2.py
|
7632f91429
fix: InternLM2 model with Tensor Parallel (#980)
|
hai 2 meses |
|
internvl.py
|
91d03c04d2
VLM: refactor composite weight loading logic (#1097)
|
hai 1 mes |
|
jais.py
|
0dfa6b60ec
core: support logprobs with multi-step scheduling (#963)
|
hai 2 meses |
|
jamba.py
|
135dfd648b
fix: LoRA support for Cohere and Jamba models (#1004)
|
hai 2 meses |
|
llama.py
|
0dfa6b60ec
core: support logprobs with multi-step scheduling (#963)
|
hai 2 meses |
|
llama_embedding.py
|
9022c6d869
remove progress_bar imports
|
hai 4 meses |
|
llava.py
|
91d03c04d2
VLM: refactor composite weight loading logic (#1097)
|
hai 1 mes |
|
llava_next.py
|
7ce3174039
VLM: refactor blip models to support composite weight loading (#1098)
|
hai 1 mes |
|
llava_next_video.py
|
91d03c04d2
VLM: refactor composite weight loading logic (#1097)
|
hai 1 mes |
|
llava_onevision.py
|
a5bfc2bc3d
VLM: add support for LLaVA-Onevision model (#1100)
|
hai 1 mes |
|
mamba.py
|
3bb0f07461
chore: rename `task_handler` to `worker` (#985)
|
hai 2 meses |
|
mamba_cache.py
|
a113309876
kernel: add meta functions for ops to prevent graph breaks (#1019)
|
hai 2 meses |
|
medusa.py
|
0dfa6b60ec
core: support logprobs with multi-step scheduling (#963)
|
hai 2 meses |
|
minicpm.py
|
ce7b602f03
model: add support for MiniCPM-3 (#1044)
|
hai 2 meses |
|
minicpm3.py
|
ce7b602f03
model: add support for MiniCPM-3 (#1044)
|
hai 2 meses |
|
minicpmv.py
|
651678d2df
VLM: use `SequenceData.from_token_counts` to create dummy data (#1093)
|
hai 1 mes |
|
mixtral.py
|
0dfa6b60ec
core: support logprobs with multi-step scheduling (#963)
|
hai 2 meses |
|
mixtral_quant.py
|
fa84f8102e
kernels: split marlin kernels for faster compile, fix MoE, temporarily remove HQQ (#1119)
|
hai 1 mes |
|
mlp_speculator.py
|
0dfa6b60ec
core: support logprobs with multi-step scheduling (#963)
|
hai 2 meses |
|
molmo.py
|
acc0c727c8
vlm: add support for molmo vision model (#1069)
|
hai 2 meses |
|
mpt.py
|
0dfa6b60ec
core: support logprobs with multi-step scheduling (#963)
|
hai 2 meses |
|
na_vit.py
|
9f3e7c86e2
feat: add fused Marlin MoE kernel (#934)
|
hai 2 meses |
|
nemotron.py
|
0dfa6b60ec
core: support logprobs with multi-step scheduling (#963)
|
hai 2 meses |
|
olmo.py
|
0dfa6b60ec
core: support logprobs with multi-step scheduling (#963)
|
hai 2 meses |
|
olmoe.py
|
0dfa6b60ec
core: support logprobs with multi-step scheduling (#963)
|
hai 2 meses |
|
opt.py
|
0dfa6b60ec
core: support logprobs with multi-step scheduling (#963)
|
hai 2 meses |
|
orion.py
|
0dfa6b60ec
core: support logprobs with multi-step scheduling (#963)
|
hai 2 meses |
|
paligemma.py
|
91d03c04d2
VLM: refactor composite weight loading logic (#1097)
|
hai 1 mes |
|
persimmon.py
|
0dfa6b60ec
core: support logprobs with multi-step scheduling (#963)
|
hai 2 meses |
|
phi.py
|
0dfa6b60ec
core: support logprobs with multi-step scheduling (#963)
|
hai 2 meses |
|
phi3.py
|
ec17b6c4d0
fix: Phi3.5 Mini and MoE LoRA inference (#1070)
|
hai 2 meses |
|
phi3_small.py
|
0dfa6b60ec
core: support logprobs with multi-step scheduling (#963)
|
hai 2 meses |
|
phi3v.py
|
80be38ca6f
chore: expose phi3_v num_crops as an mm_processor_kwargs (#1117)
|
hai 1 mes |
|
phimoe.py
|
ec17b6c4d0
fix: Phi3.5 Mini and MoE LoRA inference (#1070)
|
hai 2 meses |
|
pixtral.py
|
651678d2df
VLM: use `SequenceData.from_token_counts` to create dummy data (#1093)
|
hai 1 mes |
|
qwen.py
|
651678d2df
VLM: use `SequenceData.from_token_counts` to create dummy data (#1093)
|
hai 1 mes |
|
qwen2.py
|
c90abcc603
VLM: add pipeline parallelism support for Qwen2-VL (#1103)
|
hai 1 mes |
|
qwen2_moe.py
|
5224389dae
chore: skip loading extra bias for qwen2 moe GPTQ (#1011)
|
hai 2 meses |
|
qwen2_vl.py
|
c90abcc603
VLM: add pipeline parallelism support for Qwen2-VL (#1103)
|
hai 1 mes |
|
siglip.py
|
a5bfc2bc3d
VLM: add support for LLaVA-Onevision model (#1100)
|
hai 1 mes |
|
solar.py
|
0dfa6b60ec
core: support logprobs with multi-step scheduling (#963)
|
hai 2 meses |
|
stablelm.py
|
0dfa6b60ec
core: support logprobs with multi-step scheduling (#963)
|
hai 2 meses |
|
starcoder2.py
|
0dfa6b60ec
core: support logprobs with multi-step scheduling (#963)
|
hai 2 meses |
|
ultravox.py
|
18e0b0e932
chore: support loading weights by ID within models (#1116)
|
hai 1 mes |
|
utils.py
|
91d03c04d2
VLM: refactor composite weight loading logic (#1097)
|
hai 1 mes |
|
xverse.py
|
0dfa6b60ec
core: support logprobs with multi-step scheduling (#963)
|
hai 2 meses |
AlpinDale
fa84f8102e
kernels: split marlin kernels for faster compile, fix MoE, temporarily remove HQQ (#1119)