Commit History
add peft install back since it doesn't get installed by setup.py (#331)
db2a358
unverified
update prompts for open orca to match the paper (#317)
3d4984b
unverified
Merge pull request #307 from OpenAccess-AI-Collective/xgen-user-sharegpt-tokens
40a53ff
unverified
Merge pull request #313 from OpenAccess-AI-Collective/tokenizer-llama2-embeddings
3ffb018
unverified
don't resize embeddings to multiples of 32x by default
1066751
better handling since xgen tokenizer breaks with convert_tokens_to_ids
2a428e8
flash attention 2
9b790d3
fix sdp attention to use the flash/mem-efficient context manaager
a032c9f
feat: use multi-core
45ac7c4
fix axolotl training args dataclass annotation
ebaec3c
misc fixes
d75adb9
Merge pull request #276 from theobjectivedad/logging_enhancement
6f16c45
unverified
Fixed pre-commit problems, fixed small bug in logging_config to handle LOG_LEVEL env var
b1f4f7a
Merge branch 'OpenAccess-AI-Collective:main' into logging_enhancement
83237b8
unverified
The Objective Dad
commited on
Add ability to pass 'name' argument to load_dataset
88089e8
Merge pull request #274 from OpenAccess-AI-Collective/NanoCode012-patch-2
168a7a0
unverified
Update log message format, IMO this is easier to read.
9234b75
Adding logging enhancement
553a86b
Feat: Add save_safetensors
5491278
Set push to hub as private by default
1514739
unverified
support for loading a model by git revision
69a2350
Merge branch 'main' into quadratic-warmup
c4cf567
unverified
better configuration for quadratic warmup
c49729d
params are adam_*, not adamw_*
19cf0bd
skip explicit model type too if using trust_remote_code
d69da99
don't use llama if trust_remote_code is set since that needs to use AutoModel path
66afb76
Merge pull request #221 from utensil/local_dataset
b9b7d4c
unverified
Fix future deprecation push_to_hub_model_id
e79c8e6
Merge pull request #255 from OpenAccess-AI-Collective/open-orca-prompts
1e5014a
unverified
Merge pull request #246 from OpenAccess-AI-Collective/sys-prompts-instruct
4066c78
unverified
open orca support
78a1e1f
Fix typing list
77bdb7d
unverified
add option for instruct w sys prompts
924bbfd
Merge pull request #224 from OpenAccess-AI-Collective/system-prompt-data
f150c02
unverified
push intermediate model checkpoints to hub
612aabd
skip the system prompt
05ab909
pylint for duplicated code for system prompts
7b57ed7
add tests and supoort for loader for sys prompt data
3a38271
initial wip to get sys prompt from dataset
8d20e0a
optionally define whether to use_fast tokenizer
47d601f
Support loading data files from a local directory
9bdd30c
add validation and tests for adamw hyperparam
cb9d3af
support adamw and grad norm hyperparams
6d0ee4b
Merge pull request #212 from OpenAccess-AI-Collective/doc-20230615-v1
a81f52d
unverified
Merge pull request #214 from OpenAccess-AI-Collective/fix-tokenizing-labels
1925eaf
unverified
add float16 docs and tweak typehints
88e17ff
bugfix for potential off by one
7925ddc
style correction
136522f
maciej.karasek
commited on
issue #205 bugfix
556fe40
maciej.karasek
commited on