Fix typo (#1231) [skip ci]
8608d80
unverified
xhedit
commited on
Commit History
Peft lotfq (#1222)
4cb7900
unverified
Revert "run PR e2e docker CI tests in Modal" (#1220) [skip ci]
8da1633
unverified
run PR e2e docker CI tests in Modal (#1217) [skip ci]
36d053f
unverified
more checks and fixes for deepspeed and fsdp (#1208) [skip ci]
e923e62
unverified
Feat/chatml add system message (#1117)
98b4762
unverified
fix(log): improve warning to clarify that lora_modules_to_save expect a list (#1197)
08719b9
unverified
Mixtral fixes 20240124 (#1192) [skip ci]
54d2ac1
unverified
Phi2 multipack (#1173)
814aee6
unverified
Falcon embeddings (#1149) [skip docker]
e799e08
unverified
jupyter lab fixes (#1139) [skip ci]
eaaeefc
unverified
Qwen2 (#1166)
f5a828a
unverified
make sure the model config loader respects the model_revision too (#1160) [skip-ci]
fccb542
unverified
Deprecate max packed sequence len (#1141)
2ce5c0d
unverified
Multipack simplify for Mixtral (#1142)
6910e6a
unverified
Add `layers_to_transform` for `lora_config` (#1118)
8487b97
unverified
xzuyn
commited on
keep gate in fp32 for 16 bit loras (#1105)
da97285
unverified
add gptneox embeddings, fix phi2 inputs, also fix the casting (#1083)
78c5b19
unverified
be more robust about checking embedding modules for lora finetunes (#1074) [skip ci]
0f10080
unverified
fix: torch_dtype mistral default to fp32 (#1050)
c3e8165
unverified
Phi2 rewrite (#1058)
732851f
unverified
feature: better device mapping for large models (#918)
bdfefaf
unverified
RL/DPO (#935)
f243c21
bump transformers and update attention class map name (#1023)
bcc78d8
unverified
Adds chat templates (#1022)
f8ae59b
unverified
remove landmark attn and xpos rope implementations (#1010)
70b46ca
unverified
Feat: Warns to add to modules_to_save when adding tokens or switching special_tokens (#787)
1ffa386
unverified
Fix Deepspeed loading (#950)
5ea3aa3
unverified
Flash attn hotfix (#951)
f1f60cb
unverified
Mixtral official (#942)
7fabc4d
unverified
Mixtral multipack (#928)
68b227a
unverified
support for mamba (#915)
40a6362
unverified
fix(tokenizer): handle fast tokenizer properly for bos/eos (#914)
fde091c
unverified
Support device_map=sequential & max_memory config parameters (#903)
992e742
unverified
fix for qwen w lora (#906)
3e3229e
unverified
Feat: Add Qwen (#894)
1115c50
unverified
Phi update 202311 (#876)
9bf854e
unverified
allow overriding of model_config parameters from the YML (#853)
1bc1186
unverified
fix model parallel (#816)
964d858
unverified
fix(tokenizer): update log order after update (#806)
10388a8
unverified
fix(config): Set eos/bos to tokenizer if different (#801)
637ed09
unverified
refactor neft patch to be more re-usable similar to trl's impl (#796)
827ec3d
unverified
chore: refactor truthy check and fix mypy (#780)
11d1d60
unverified
Implement fused modules (#747)
15d3a65
unverified
Fix(model): Linear detected and added to target module with rope linear (#738)
440c3ab
unverified
add noisy embedding (#721)
3bd9528
unverified
Maxime
Maxime
commited on