Skip to content

Issues: axolotl-ai-cloud/axolotl

Beta
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

165 Open 623 Closed
165 Open 623 Closed
Author
Filter by author
Loading
Label
Filter by label
Loading
Use alt + click/return to exclude labels
or ⇧ + click/return for logical OR
Projects
Filter by project
Loading
Milestones
Filter by milestone
Loading
Assignee
Filter by who’s assigned
Sort
Sort by
Newest Oldest Most commented Least commented Recently updated Least recently updated Best match
Most reactions
👍 👎 😄 🎉 😕 ❤️ 🚀 👀

Issues list

Make Axolotl Print Dataset Name Before Processing for Easy Debugging enhancement New feature or request
#2667 opened May 12, 2025 by xzuyn
5 tasks done
1
2
Qwen3 Does Not Work With fix_untrained_tokens bug Something isn't working
#2660 opened May 10, 2025 by xzuyn
6 of 8 tasks
1
2
GRPO + SP + Liger enhancement New feature or request
#2659 opened May 10, 2025 by djsaunde
5 tasks done
@djsaunde
Allow passing args to apply_chat_template enhancement New feature or request
#2658 opened May 9, 2025 by BitPhinix
5 tasks done
1
Installation error bug Something isn't working
#2652 opened May 8, 2025 by Ruiwen505
6 of 8 tasks
2
Deprecated flash_attn_fuse_qkv toggle for the latest transformers (4.51.3 or lower) enhancement New feature or request
#2650 opened May 8, 2025 by DaizeDong
5 tasks done
1
Liger + PEFT + GRPO enhancement New feature or request
#2648 opened May 7, 2025 by SalmanMohammadi
5 tasks done
Axolotl v0.10.0
@SalmanMohammadi
Axolotl complains the mapper for user_defined data type of dpo training has an issue bug Something isn't working
#2645 opened May 7, 2025 by seungduk-yanolja
5 of 8 tasks
2
Training finishes but errors out while trying to write out final LoRA weights (unable to resume) bug Something isn't working
#2633 opened May 5, 2025 by chimezie
5 of 8 tasks
2
Deepspeed Checkpoints Fail To Resume With PyTorch >=2.6 bug Something isn't working
#2610 opened May 1, 2025 by alexdremov
8 tasks done
10
FlashAttention V3 enhancement New feature or request
#2594 opened Apr 30, 2025 by casper-hansen
5 tasks done
4
Resuming a checkpoint with LORA
#2568 opened Apr 25, 2025 by RevanthRameshkumar
3 tasks done
2
Running inference of QLoRA fine-tuned Llama-4 Scout does not work bug Something isn't working
#2550 opened Apr 22, 2025 by cyril-k
6 of 8 tasks
6
QAT - Quantization Aware Training enhancement New feature or request
#2540 opened Apr 18, 2025 by bratao
5 tasks done
2
GRPO on Modal - vllm doesn't start bug Something isn't working
#2528 opened Apr 13, 2025 by floleuerer
5 of 8 tasks
@SalmanMohammadi
3
Trion 3.2.0 Doesn't Work with GRPO+vllm bug Something isn't working
#2526 opened Apr 12, 2025 by RevanthRameshkumar
6 of 8 tasks
6
Corda Config Error
#2507 opened Apr 9, 2025 by aitechguy0105
3 tasks done
1
Support DeepSpeed AutoTP enhancement New feature or request
#2491 opened Apr 5, 2025 by casper-hansen
5 tasks done
1
3
Add support for PEFT trainable tokens enhancement New feature or request
#2431 opened Mar 21, 2025 by NanoCode012
5 tasks done
illegal memory access when using pretraining_dataset: , works when swapped to datasets bug Something isn't working
#2418 opened Mar 17, 2025 by lzumot
6 of 8 tasks
2
DPO with the chat_template strategy fails due to extra columns being passed bug Something isn't working
#2415 opened Mar 14, 2025 by lhl
6 of 8 tasks
1
"CUDA error: invalid argument" with FSDP + QLora finetuning bug Something isn't working
#2409 opened Mar 12, 2025 by peterwilli
6 of 8 tasks
11
Process hanged when using cpu offloading bug Something isn't working waiting for reporter
#2405 opened Mar 11, 2025 by mohit-217
5 of 8 tasks
2
FutureWarning: FSDP.state_dict_type() and FSDP.set_state_dict_type() are being deprecated. Please use APIs, get_state_dict() and set_state_dict(), which can support different parallelisms, FSDP1, FSDP2, DDP. enhancement New feature or request
#2402 opened Mar 11, 2025 by NanoCode012
5 tasks done
1
EXTREMELY SLOW (unusable) towards end of tokenization of dataset with long multi turn conversations bug Something isn't working
#2396 opened Mar 7, 2025 by Nero10578
6 of 8 tasks
8
ProTip! Exclude everything labeled bug with -label:bug.