add sdpa to OPT #33298
ArthurZucker
merged
57 commits into
huggingface:main
from
avishaiElmakies:spda_opt
add sdpa to OPT
35e1a643
chore: remove redundant whitespace in OPTDecoder class
908e39bd
fixup
c84a4dd1
bug fix
be32f920
add sdpa and attention generate test
80639948
fixup
248029a6
Refactor OPTAttention forward method for improved readability and maiā¦
b66e3d85
undo refactor for _shape and key,val states
579d60e8
add OPT to doc, fixup didn't find it for some reason
b1053765
change order
c3496328
change default attn_implemntation in testing to eager
6dba8b00
Merge branch 'main' into spda_opt
989625ba
[run-slow] opt
1d21751f
change test_eager_matches_sdpa_generate to the one llama
7233fda5
Update default attention implementation in testing common
9bacdeb2
[run-slow] opt
5b38f784
remove uneeded print
3f24a047
[run-slow] opt
2efd25aa
refactor model testers to have attn_implementation="eager"
bdd9cb23
[run-slow] opt
f80e3b3c
vasqu
commented
on 2024-09-21
convert test_eager_matches_sdpa_generate to opt-350M
7ea22eb7
bug fix when creating mask for opt
b5547e7c
Merge branch 'main' into spda_opt
eaa80282
[run-slow] opt
668e291c
if layer head mask default to eager
d9d3bb3a
if head mask is not none fall to eager
388d663f
[run-slow] opt
e735ec4b
vasqu
commented
on 2024-09-22
amyeroberts
commented
on 2024-09-25
Update src/transformers/models/opt/modeling_opt.py
f94d5742
Clean up Unpack imports (#33631)
e734d9d1
Fix DPT /Dinov2 sdpa regression on main (#33660)
34593ba9
handle dependency errors in check_imports (#33622)
6889d696
add back self.max_position_embeddings = config.max_position_embeddingā¦
d488c33b
Fix Llava conversion for LlavaQwen2ForCausalLM with Clip vision towerā¦
99909159
Uniformize kwargs for Udop processor and update docs (#33628)
3720ecab
Generation: deprecate `PreTrainedModel` inheriting from `GenerationMiā¦
9b11d289
Enable BNB multi-backend support (#31098)
d3f8417e
Fix error string after refactoring into get_chat_template (#33652)
52a0a757
uniformize git processor (#33668)
400927e7
Modular `transformers`: modularity and inheritance for new model addiā¦
3b0d24c8
Fix CIs post merging modular transformers (#33681)
ef64c818
Fixed docstring for cohere model regarding unavailability of prune_heā¦
6cd88aae
Generation tests: update imagegpt input name, remove unused functionsā¦
4a457c18
Improve Error Messaging for Flash Attention 2 on CPU (#33655)
4deac160
Gemma2: fix config initialization (`cache_implementation`) (#33684)
1f7d50a0
Fix ByteLevel alphabet missing when Sequence pretokenizer is used (#3ā¦
3e798fa9
Uniformize kwargs for image-text-to-text processors (#32544)
9665ecca
šØšØ Setting default behavior of assisted decoding (#33657)
e1839b97
tests: fix pytorch tensor placement errors (#33485)
37da2d61
bump tokenizers, fix added tokens fast (#32535)
58c2b2b7
[Pixtral] Improve docs, rename model (#33491)
f0bb0a8f
fix code quality after merge
34a9142c
HFQuantizer implementation for compressed-tensors library (#31704)
6aeec65b
Merge branch 'main' into spda_opt
3e69375e
update model card for opt
a9b18dc9
add batch size to inference table
9876dbba
[slow-run] opt
ff35bbc2
[run-slow] opt
cfd1209a
Login to write a write a comment.
Login via GitHub