| .. |
|
__init__.py
|
6b1fdd07bd
chore: add isort and refactor formatting script and utils
|
пре 4 месеци |
|
arctic.py
|
0f4a9ee77b
quantized lm_head (#582)
|
пре 5 месеци |
|
baichuan.py
|
0f4a9ee77b
quantized lm_head (#582)
|
пре 5 месеци |
|
blip.py
|
9f010deb8a
feat: add blip-2 support
|
пре 4 месеци |
|
blip2.py
|
9f010deb8a
feat: add blip-2 support
|
пре 4 месеци |
|
bloom.py
|
4d4e767838
ci: take one of fixing lint issues
|
пре 4 месеци |
|
chameleon.py
|
c9310eeb02
fix: skip loading lm_head for tie_word_embeddings models
|
пре 4 месеци |
|
chatglm.py
|
0f4a9ee77b
quantized lm_head (#582)
|
пре 5 месеци |
|
clip.py
|
e26a4ac698
chore: avoid loading the unused layers and init the VLM up to the required feature space
|
пре 5 месеци |
|
commandr.py
|
0f4a9ee77b
quantized lm_head (#582)
|
пре 5 месеци |
|
dbrx.py
|
0f4a9ee77b
quantized lm_head (#582)
|
пре 5 месеци |
|
decilm.py
|
56e0b8223c
chore: add base class for LoRA-supported models
|
пре 5 месеци |
|
deepseek.py
|
0f4a9ee77b
quantized lm_head (#582)
|
пре 5 месеци |
|
deepseek_v2.py
|
1efd0f89b7
feat: support FP8 for DeepSeekV2 MoE
|
пре 5 месеци |
|
falcon.py
|
4d4e767838
ci: take one of fixing lint issues
|
пре 4 месеци |
|
fuyu.py
|
9e9515f39a
fix: feature size calculation for Llava-next
|
пре 4 месеци |
|
gemma.py
|
4d4e767838
ci: take one of fixing lint issues
|
пре 4 месеци |
|
gemma2.py
|
4d4e767838
ci: take one of fixing lint issues
|
пре 4 месеци |
|
gpt2.py
|
4d4e767838
ci: take one of fixing lint issues
|
пре 4 месеци |
|
gpt_bigcode.py
|
4d4e767838
ci: take one of fixing lint issues
|
пре 4 месеци |
|
gpt_j.py
|
0f4a9ee77b
quantized lm_head (#582)
|
пре 5 месеци |
|
gpt_neox.py
|
0f4a9ee77b
quantized lm_head (#582)
|
пре 5 месеци |
|
idefics2_vision_model.py
|
9a50e3b4eb
refactor: minicpmv and port Idefix2VisionTransformer
|
пре 4 месеци |
|
interfaces.py
|
e76bbe72eb
chore: handle aborted requests for jamba
|
пре 5 месеци |
|
intern_vit.py
|
92987963a4
fix: RMSNorm forward in InternViT attention qk_layernorm
|
пре 4 месеци |
|
internlm2.py
|
9cf1275f03
feat: add internvl support
|
пре 4 месеци |
|
internvl.py
|
9e9515f39a
fix: feature size calculation for Llava-next
|
пре 4 месеци |
|
jais.py
|
4d4e767838
ci: take one of fixing lint issues
|
пре 4 месеци |
|
jamba.py
|
f83eb07fd1
feat: use FusedMoE for jamba
|
пре 4 месеци |
|
llama.py
|
4d4e767838
ci: take one of fixing lint issues
|
пре 4 месеци |
|
llama_embedding.py
|
50b7c13db0
refactor: attention selector (#552)
|
пре 6 месеци |
|
llava.py
|
acbdc50a71
fix: `vocab_size` field access in llava
|
пре 5 месеци |
|
llava_next.py
|
9e9515f39a
fix: feature size calculation for Llava-next
|
пре 4 месеци |
|
medusa.py
|
d9f4c36edd
feat: Medusa speculative decoding support (#590)
|
пре 5 месеци |
|
minicpm.py
|
c9310eeb02
fix: skip loading lm_head for tie_word_embeddings models
|
пре 4 месеци |
|
minicpmv.py
|
9a50e3b4eb
refactor: minicpmv and port Idefix2VisionTransformer
|
пре 4 месеци |
|
mixtral.py
|
00503b9fc1
feat: non-uniform quantization via `compressed-tensors` for llama
|
пре 5 месеци |
|
mixtral_quant.py
|
0f4a9ee77b
quantized lm_head (#582)
|
пре 5 месеци |
|
mlp_speculator.py
|
db73f03cdc
fix: use ParallelLMHead for MLPSpeculator
|
пре 5 месеци |
|
mpt.py
|
4d4e767838
ci: take one of fixing lint issues
|
пре 4 месеци |
|
na_vit.py
|
9a50e3b4eb
refactor: minicpmv and port Idefix2VisionTransformer
|
пре 4 месеци |
|
nemotron.py
|
18b45266bb
feat: add nemotron HF support (#606)
|
пре 4 месеци |
|
olmo.py
|
c9310eeb02
fix: skip loading lm_head for tie_word_embeddings models
|
пре 4 месеци |
|
opt.py
|
4d4e767838
ci: take one of fixing lint issues
|
пре 4 месеци |
|
orion.py
|
0f4a9ee77b
quantized lm_head (#582)
|
пре 5 месеци |
|
paligemma.py
|
c3ee71a437
feat: port SiglipVisionModel from transformers
|
пре 4 месеци |
|
persimmon.py
|
e13a66925c
feat: add fuyu vision model and persimmon language model support
|
пре 5 месеци |
|
phi.py
|
0f4a9ee77b
quantized lm_head (#582)
|
пре 5 месеци |
|
phi3_small.py
|
0f4a9ee77b
quantized lm_head (#582)
|
пре 5 месеци |
|
phi3v.py
|
9e9515f39a
fix: feature size calculation for Llava-next
|
пре 4 месеци |
|
qwen.py
|
0f4a9ee77b
quantized lm_head (#582)
|
пре 5 месеци |
|
qwen2.py
|
d357341203
chore: add pipeline parallel support for Qwen
|
пре 4 месеци |
|
qwen2_moe.py
|
d357341203
chore: add pipeline parallel support for Qwen
|
пре 4 месеци |
|
siglip.py
|
4d4e767838
ci: take one of fixing lint issues
|
пре 4 месеци |
|
stablelm.py
|
0f4a9ee77b
quantized lm_head (#582)
|
пре 5 месеци |
|
starcoder2.py
|
0f4a9ee77b
quantized lm_head (#582)
|
пре 5 месеци |
|
utils.py
|
165a3aa7b3
fix: fp8 marlin and cpu offloading with fp8 marlin
|
пре 4 месеци |
|
xverse.py
|
0f4a9ee77b
quantized lm_head (#582)
|
пре 5 месеци |
AlpinDale
4d4e767838
ci: take one of fixing lint issues