Spaces:

Dovakiins
/

qwerrwe

Build error

App Files Files Community

qwerrwe / src /axolotl /monkeypatch

100 contributors

History: 63 commits

winglian's picture

allow the optimizer prune ratio for ReLoRA to be configurable (#1287)

4b997c3 unverified 12 months ago

data
support for true batches with multipack (#1230) about 1 year ago
mixtral
simplify haldning for newer multipack patches so they can be added in a single place (#1270) 12 months ago
btlm_attn_hijack_flash.py

2.32 kB

flash_attention + sample packing for stablelm 3b (#671) over 1 year ago
fastchat_conversation_turns.py

8.33 kB

Added chatglm3 conversation type for training models like TinyLLama (#1036) about 1 year ago
llama_attn_hijack_flash.py

32 kB

Add shifted sparse attention (#973) [skip-ci] about 1 year ago
llama_attn_hijack_xformers.py

5.69 kB

various bugfixes (#856) about 1 year ago
llama_expand_mask.py

672 Bytes

support for true batches with multipack (#1230) about 1 year ago
llama_patch_multipack.py

1.1 kB

support for true batches with multipack (#1230) about 1 year ago
mistral_attn_hijack_flash.py

22.5 kB

Respect sliding_window=None (#1214) about 1 year ago
multipack.py

1.16 kB

simplify haldning for newer multipack patches so they can be added in a single place (#1270) 12 months ago
relora.py

16.8 kB

allow the optimizer prune ratio for ReLoRA to be configurable (#1287) 12 months ago
stablelm_attn_hijack_flash.py

15.4 kB

flash_attention + sample packing for stablelm 3b (#671) over 1 year ago
utils.py

8.05 kB

Add MPS support (#1264) 12 months ago