eole-nlp eole issues - Githubissues

eole-nlp / eole

Open language modeling toolkit based on PyTorch

https://eole-nlp.github.io/eole

MIT License

15 stars 5 forks source link

issues

Newest

Newest Most commented Recently updated Oldest Least commented Least recently updated

Refac position encoding

#60 vince62s opened 3 hours ago
0
The unknown token in the vocabulary of finetuned checkpoint breaks the prediction at inference time

#59 l-k-11235 opened 1 day ago
7
Fix prefix and suffix transforms - avoid adding empty suffix or prefix

#57 sersh88 opened 2 days ago
1
prefix transform bug?

#56 sersh88 opened 4 days ago
3
bfloat16 support, and an attempt at homogenizing model_dtype & precision

#54 francoishernandez opened 6 days ago
3
Add Recipe to train a cometkiwi-like encoder model (which can be used to score sentence pairs)

#53 vince62s closed 5 days ago
0
Simplify __init__ files, remove some unused code

#52 francoishernandez closed 4 days ago
0
[fix] Fix paths in wiki_103 recipe, add pyarrow opt requirement

#51 francoishernandez closed 1 week ago
0
Enable PyPI release workflow

#50 francoishernandez closed 1 week ago
0
[fix] Allow to build_vocab with full train config, patch vocab validation

#49 francoishernandez closed 1 week ago
0
Can't follow recipe

#48 Kai-Piontek closed 5 days ago
2
[docs] Github Actions workflow to facilitate docs deployment

#47 francoishernandez closed 1 week ago
0
Estim first token instead of average

#46 vince62s closed 1 week ago
0
[WIP] Rework handling of special tokens

#45 francoishernandez opened 1 week ago
0
Some improvements to config.json readability

#44 francoishernandez closed 1 week ago
0
Some fixes, get rid of data_task, homogenize model_task to model_type

#43 francoishernandez closed 1 week ago
0
[WIP] Inference server, lots of related changes

#42 francoishernandez opened 2 weeks ago
0
Add support for XLM-Roberta-XL (and XXL) conversion

#41 vince62s closed 2 weeks ago
1
estim lambda scheduler

#40 vince62s closed 2 weeks ago
0
Forgot hellaswag.py tool in #38

#39 francoishernandez closed 2 weeks ago
0
Add gpt2 converter, hellaswag eval tool, misc fixes

#38 francoishernandez closed 2 weeks ago
0
[patch] upgrade docusaurus deps, fix build script

#37 francoishernandez closed 2 weeks ago
0
Resize the key_pad_mask

#36 l-k-11235 closed 3 weeks ago
2
Encoder only work

#35 vince62s closed 1 week ago
0
COMET scoring

#34 vince62s opened 3 weeks ago
0
revamp default space tokenization - review the ((newline)) thing

#33 vince62s opened 3 weeks ago
0
[WIP] fineweb10B/gpt2 recipe, and supporting changes

#32 francoishernandez opened 3 weeks ago
0
Fine-tuning fails with error AssertionError: An error in model's partition and checkpoint's slice was detected

#31 randy-ac closed 3 weeks ago
4
fix missing layers names

#30 vince62s closed 3 weeks ago
0
Split MHA

#29 vince62s closed 3 weeks ago
0
[fix] Patch lora bin to dump json config

#28 francoishernandez closed 3 weeks ago
0
fix filtertoolong transform when there is an empty token

#27 vince62s closed 3 weeks ago
0
Refacto convert HF

#26 funboarder13920 opened 3 weeks ago
2
review flash/sdpa arg

#25 vince62s closed 3 weeks ago
0
`config.models.BaseModelConfig._override_values` updates everything once

#24 francoishernandez closed 3 weeks ago
0
missing removal of average attn

#23 vince62s closed 3 weeks ago
0
Revert "MHA refac: rope without complex operations + query only as input of the forward"

#22 vince62s closed 3 weeks ago
0
change self_attn_type

#21 vince62s closed 3 weeks ago
0
MHA refac: rope without complex operations + query only as input of the forward

#20 vince62s closed 4 weeks ago
0
Issue in validation order with model level parameters propagation

#19 funboarder13920 closed 3 weeks ago
1
The finetuning in tensor parallel mode does not work as expected

#18 l-k-11235 closed 3 weeks ago
4
refactor position encoding settings

#17 vince62s opened 4 weeks ago
0
refactor Rope interleave=True mode to avoid using Complex/Polar operations

#16 vince62s closed 4 weeks ago
1
remove unsused average attn

#15 vince62s closed 4 weeks ago
0
Fix the tokenizer saving in the HF converter

#14 l-k-11235 closed 4 weeks ago
0
fix mmlu config

#13 vince62s closed 4 weeks ago
0
rename num_kv remove multiquery

#12 vince62s closed 1 month ago
0
`eole model lora` fails to save the config

#11 l-k-11235 closed 3 weeks ago
4
Fix the checkpoint directory cleaning

#10 l-k-11235 closed 1 month ago
0
The `cleanup`method of the `TrainingModelSaver` returns `FileNotFoundError`

#9 l-k-11235 closed 1 month ago
0