OpenAccess-AI-Collective axolotl issues

OpenAccess-AI-Collective / axolotl

Go ahead and axolotl questions

https://openaccess-ai-collective.github.io/axolotl/

Apache License 2.0

6.4k stars 714 forks source link

issues

Newest

Newest Most commented Recently updated Oldest Least commented Least recently updated

Fix the broken link in README

#1678 saeedesmaili opened 23 hours ago
0
need to add back drop_last for sampler

#1676 winglian closed 1 day ago
0
cleanup the deepspeed proxy model at the end of training

#1675 winglian closed 2 days ago
0
use mixins for orpo and kto configs so they work with axolotl customi zations

#1674 winglian closed 3 days ago
0
use L4 vs A10G for modal cicd

#1673 winglian opened 3 days ago
0
revert multipack batch sampler changes

#1672 winglian closed 3 days ago
1
handle the system role too for chat templates

#1671 winglian closed 3 days ago
0
Train FAILED. Crashed while training with SIGTERM

#1670 RodriMora opened 4 days ago
0
make sure the CI fails when pytest script fails

#1669 winglian closed 3 days ago
0
Fix README quick start example usage model dirs

#1668 abevoelker closed 4 days ago
0
Correct name of MixtralBlockSparseTop2MLP (L -> l)

#1667 seungduk-yanolja closed 4 days ago
0
Llama Reserved Tokens Initialization

#1666 cinjon opened 4 days ago
0
fix lint issue that snuck through

#1665 winglian closed 4 days ago
0
set chat_template in datasets config automatically

#1664 winglian closed 3 days ago
0
update deps

#1663 winglian closed 4 days ago
0
Fix Google Colab notebook 2024-05

#1662 maciejgryka closed 4 days ago
0
Not enough information on pre-tokenized dataset.

#1661 rumbleFTW opened 5 days ago
0
Generalizing the chat_template prompt strategy

#1660 fozziethebeat closed 4 days ago
1
Fix Lora config error for Llama3

#1659 oaishi closed 4 days ago
0
Pulling the image from Docker retuns with ''unauthorized: authentication required''

#1658 Fischherboot opened 1 week ago
0
Fix setting correct repo id when pushing dataset to hub

#1657 chrislee973 opened 1 week ago
0
Fix tokenization for CodeQwen models

#1656 artemdinaburg closed 1 week ago
3
Fix: ensure correct handling of `val_set_size` as `float` or `int`

#1655 davidecaroselli closed 4 days ago
0
Generalize the `chat_template` prompt strategy with more configuration options

#1654 fozziethebeat closed 4 days ago
5
document how to use `share_strategy="no"`

#1653 charlesfrye closed 1 week ago
0
load explicit splits on datasets

#1652 winglian closed 3 days ago
0
support for custom messages field in sharegpt

#1651 winglian closed 1 week ago
0
Llama3 Lora training fails to output and save

#1650 austinm1120 opened 1 week ago
0
dataset type sharegpt no longer works

#1649 thepowerfuldeez opened 1 week ago
8
Compatibility with huggingface-pytorch-training:2.0.0-transformers4.28.1-gpu-py310-cu118-ubuntu20.04 docker image

#1648 tleyden closed 1 week ago
1
allow report_to for multiple providers

#1647 winglian closed 1 week ago
0
Enable LoRA+ setting for dpo trainer

#1646 thepowerfuldeez closed 1 week ago
0
DPO Prompt Strategies only support single-turn and will fail silently on multi-turn datasets

#1645 bjoernpl opened 1 week ago
1
Llama 3 & Mistral LoRA Examples Error (needs `eval_sample_packing: False`)

#1644 VelocityRa opened 1 week ago
0
fixes to save on fractional save_steps

#1643 winglian closed 1 week ago
0
A saves_per_epoch that creates a fraction may create training that don´t have checkpoints

#1642 bratao closed 1 week ago
4
Llama 3 8b OOM with GaLore on 2x A100s (Mistral 7b is fine?)

#1641 e-p-armstrong opened 2 weeks ago
4
Add KTO support

#1640 benredmond closed 1 week ago
0
[Feature]: Support for Falcon-11B model (Falcon 2)

#1639 s-smits closed 1 week ago
3
Update tiny-llama qlora.yml addressing eval packing error

#1638 jaydeepthik closed 1 week ago
0
Support RecurrentGemma

#1637 julien-blanchon opened 2 weeks ago
1
Fix llama3 chat_template (extra <|eot_id|> on last turn)

#1635 lhl closed 1 week ago
2
add save_only_model option

#1634 jquesnelle closed 2 weeks ago
0
Test phi 3 model

#1633 vinamrabenara closed 2 weeks ago
0
Error Finetuning CodeQwen 1.5 7B: ```Column 1 named token_type_ids expected length 113 but got length 112```

#1632 artemdinaburg opened 2 weeks ago
4
Auto resume from checkpoint looks for "trainer_state.json", but no file is generated

#1631 l3utterfly opened 2 weeks ago
1
fix ray install

#1630 winglian closed 2 weeks ago
0
more fixes to work with runpod + skypilot

#1629 winglian closed 2 weeks ago
0
cloud image w/o tmux

#1628 winglian closed 2 weeks ago
0
install rsync too

#1627 winglian closed 2 weeks ago
0