Commit History
fix: override use_flash_attn in lora
dc4080e
verified
jupyterjazz
commited on
Fix rotary backward
169b7fb
verified
jupyterjazz
commited on
cpu-inference (#35)
80b1ff7
verified
feat: configurable use_reentrant (#37)
09dbf45
verified
refine-codebase (#33)
2646361
verified
jupyterjazz
commited on
fix-adapter-masks (#32)
0f0bed6
verified
change rotary base (#31)
4434bf3
verified
2-adapter-tuning (#29)
95fd08c
verified
Update modeling_xlm_roberta.py
027219a
verified
jupyterjazz
commited on
2-adapter-tuning-initial-impl (#30)
2cc8472
verified
Update modeling_xlm_roberta.py
3eceb33
verified
jupyterjazz
commited on
make-lora-stateless (#25)
e860caa
verified
jupyterjazz
commited on
feat-routing (#26)
6a92924
verified
fix-rope-inference (#28)
955fea2
verified
jupyterjazz
commited on
no-flash-attention-during-inference (#22)
7ad815b
verified
jupyterjazz
commited on
Rename config to config.json
4d09ca8
verified
jupyterjazz
commited on
Create config
721e106
verified
bwang0911
commited on
Delete config.json
a60f7c0
verified
bwang0911
commited on
rope-embeddings (#20)
e3681c2
verified
jupyterjazz
commited on
alibi (#19)
ab85772
verified
lora bugfix (#16)
7c4a80c
verified
truncate-embedding-dimension (#10)
8542ad8
verified
jupyterjazz
commited on
lora-multiple-adapters (#11)
27d23b2
verified
jupyterjazz
commited on
Refactor LoRA (#8)
4b000ec
verified
jupyterjazz
commited on
support lora (#1)
f9b3adb
verified
jupyterjazz
commited on
Support for SequenceClassification (#7)
0bb73e5
verified
michael-guenther
commited on
Support torch_dtype and CLS pooling (#6)
13c4251
verified
michael-guenther
commited on
Update modeling_xlm_roberta_for_glue.py
102e7bc
verified
koukandre
commited on
add-encode-function (#5)
424df3c
verified
michael-guenther
commited on
support-cpu (#2)
290e593
verified
michael-guenther
commited on
Update configuration_xlm_roberta.py
807ba34
verified
jupyterjazz
commited on
Update configuration_xlm_roberta.py
9db6c6f
verified
jupyterjazz
commited on
add stochastic_depth
77af1c7
michael-guenther
commited on
support activation checkpointing
1c61b96
michael-guenther
commited on
add mlm model and adjust naming
95b4916
michael-guenther
commited on
add script to convert weights
eb21270
michael-guenther
commited on
add tokenizer class
3d87c79
michael-guenther
commited on
change config name
30e6a10
michael-guenther
commited on
upload model
2e3ebcb
michael-guenther
commited on
initial commit
2aec9c9
verified
michael-guenther
commited on