Ecosyste.ms: Issues
An open API service for providing issue and pull request metadata for open source projects.
GitHub / unslothai/unsloth issues and pull requests
#970 - Infinite Generation
Issue -
State: open - Opened by MuhammadBilalKhan267 3 months ago
- 5 comments
#969 - how can i compare before and after finetune model results at same time
Issue -
State: open - Opened by paras007frnd 3 months ago
- 1 comment
#968 - Ubsloth defaults to 0 GPU always
Issue -
State: open - Opened by kishoretvk 3 months ago
- 5 comments
#967 - FastLanguageModel import error
Issue -
State: closed - Opened by VatsalPatel18 3 months ago
- 1 comment
#966 - Add StepFlushCallback
Pull Request -
State: closed - Opened by vTuanpham 3 months ago
- 1 comment
#965 - Error when saving unsloth Llama 3.1 model to GGUF format
Issue -
State: open - Opened by okoliechykwuka 3 months ago
- 3 comments
#964 - ModuleNotFoundError: No module named 'triton'
Issue -
State: closed - Opened by boseong-yun 3 months ago
- 7 comments
Labels: fixed - pending confirmation, URGENT BUG
#963 - TypeError in `orpo_trainer.train()`: 'str' object is not callable
Issue -
State: open - Opened by kdunee 3 months ago
- 7 comments
Labels: fixed - pending confirmation
#962 - can't import FastLanguageModel from unsloth
Issue -
State: open - Opened by sagniknandigit 3 months ago
- 2 comments
#961 - AttributeError: 'RMSNorm' object has no attribute 'variance_epsilon'
Issue -
State: closed - Opened by Silentssss 3 months ago
- 3 comments
Labels: fixed - pending confirmation
#960 - KeyError: 'base_model.model.model.layers.0.mlp.down_proj.lora_A.weight'
Issue -
State: open - Opened by Iven2132 3 months ago
- 4 comments
#959 - Multiple Generation Similar to Huggingface `num_return_sequences`
Issue -
State: open - Opened by ankitprezent 3 months ago
- 1 comment
#959 - Multiple Generation Similar to Huggingface `num_return_sequences`
Issue -
State: open - Opened by ankitprezent 3 months ago
- 2 comments
#958 - How to fine-tune using pytroch dataset instead of hf's dataset
Issue -
State: open - Opened by xugy16 3 months ago
- 2 comments
#957 - issue while finetuning llama 3.1, 'tuple' object has no attribute 'remove_unused_columns'
Issue -
State: open - Opened by Abhisekgit1994 3 months ago
- 6 comments
#956 - Conda environment setup issues
Issue -
State: closed - Opened by timlai4 3 months ago
- 7 comments
Labels: currently fixing
#955 - Phi 3.5 bug fix
Pull Request -
State: closed - Opened by danielhanchen 3 months ago
#954 - Able to finetune `homebrewltd/llama3.1-s-instruct-v0.2` (Input=Text & Audio, Output=Text)
Issue -
State: open - Opened by asmith26 3 months ago
- 3 comments
#953 - KeyError: '__EOS_TOKEN__'
Issue -
State: closed - Opened by zs856 3 months ago
- 4 comments
Labels: currently fixing
#952 - update token retrieval logic
Pull Request -
State: closed - Opened by not-lain 3 months ago
- 2 comments
#951 - Exclamation points are not periods.
Issue -
State: open - Opened by briandw 3 months ago
- 2 comments
#950 - Output_attentions not available
Issue -
State: open - Opened by Paoloc99 3 months ago
- 2 comments
#949 - Add nvidia/Minitron-8B-Base support
Issue -
State: open - Opened by minipasila 3 months ago
- 1 comment
Labels: feature request
#948 - Your Flash Attention 2 installation seems to be broken
Issue -
State: closed - Opened by C0casio45 3 months ago
- 5 comments
#947 - Fix DPO
Pull Request -
State: closed - Opened by danielhanchen 3 months ago
#946 - AttributeError: 'LongRopeRotaryEmbedding' object has no attribute 'inv_freq' when finetuning Phi3.5 mini
Issue -
State: open - Opened by beniz 3 months ago
- 9 comments
Labels: fixed - pending confirmation
#945 - stablelm-zephyr-3b support
Issue -
State: open - Opened by ItzAmirreza 3 months ago
- 2 comments
Labels: feature request
#944 - GPU = NVIDIA GeForce RTX 4060 Ti 16G , Finetunning unsloth/Meta-Llama-3.1-8B-bnb-4bit OOM
Issue -
State: open - Opened by 1272870698 3 months ago
- 1 comment
#943 - model loading failed: RuntimeError: Error(s) in loading state_dict for PeftModelForCausalLM: size mismatch for base_model.model.model.embed_tokens.weight
Issue -
State: open - Opened by BoyangGu1 3 months ago
- 5 comments
Labels: currently fixing
#942 - Save the original model name in `config.json` instead of mapped name when saving LoRA
Issue -
State: open - Opened by ivsanro1 3 months ago
- 4 comments
Labels: currently fixing
#941 - Phi 3.5
Pull Request -
State: closed - Opened by danielhanchen 3 months ago
#940 - Phi 3.5
Pull Request -
State: closed - Opened by danielhanchen 3 months ago
#939 - 'tuple' object has no attribute 'remove_unused_columns'
Issue -
State: open - Opened by Abhisekgit1994 3 months ago
- 3 comments
#938 - Update README.md
Pull Request -
State: closed - Opened by danielhanchen 3 months ago
#937 - Fix NEFTune
Pull Request -
State: closed - Opened by danielhanchen 3 months ago
#936 - issue with Kaggle continue pretraining
Issue -
State: closed - Opened by Dusker233 3 months ago
- 7 comments
#935 - issue with Merging lora adapters
Issue -
State: closed - Opened by Ammar-Alnagar 3 months ago
- 1 comment
#934 - Unable to load unsloth trained model saved to a local directory.
Issue -
State: open - Opened by InderjeetVishnoi 3 months ago
- 7 comments
Labels: currently fixing
#933 - FlashAttention only support fp16 and bf16 data type
Issue -
State: open - Opened by ArcherShirou 3 months ago
- 3 comments
#932 - Supporting "LoRA-GA: Low-Rank Adaptation with Gradient Approximation"?
Issue -
State: open - Opened by fzyzcjy 3 months ago
- 2 comments
Labels: feature request
#931 - Bug #930
Pull Request -
State: closed - Opened by danielhanchen 3 months ago
#930 - newest commit (untrained tokens llame 3.1 base) creates bfloat16 issue with Mistral Nemo training
Issue -
State: closed - Opened by Nazzaroth2 3 months ago
- 2 comments
Labels: fixed - pending confirmation
#929 - untrained tokens llama 3.1 base
Pull Request -
State: closed - Opened by danielhanchen 3 months ago
#928 - Something is wrong with `save_pretrained_merged` or `FastLanguageModel.from_pretrained`
Issue -
State: closed - Opened by tengerye 3 months ago
- 3 comments
#927 - NameError: name 'os' is not defined
Issue -
State: open - Opened by Tizzzzy 3 months ago
- 1 comment
#926 - Issue with running trainer.train()
Issue -
State: open - Opened by dilerbatu 3 months ago
- 1 comment
#925 - Can not load tokenizer after `save_pretrained_merged`
Issue -
State: closed - Opened by tengerye 3 months ago
- 2 comments
#924 - pass trust_remote_code to AutoConfig.from_pretrained and PeftConfig .from_pretrained
Issue -
State: open - Opened by wellhowtosay 3 months ago
- 1 comment
Labels: fixed - pending confirmation
#923 - beam search does not work for gemma2b
Issue -
State: open - Opened by world2vec 3 months ago
- 5 comments
Labels: currently fixing, URGENT BUG
#922 - Jamba Models not Supported yet
Issue -
State: open - Opened by chintanckg 3 months ago
Labels: feature request
#921 - Fix mapping
Pull Request -
State: closed - Opened by danielhanchen 3 months ago
#920 - Bug Fixes
Pull Request -
State: closed - Opened by danielhanchen 3 months ago
#919 - Is it possible to use some of the kernels with deepspeed?
Issue -
State: closed - Opened by huyiwen 3 months ago
- 2 comments
#918 - Gemma 2 9b lm_head, emb_tokens probably not loading (but were saved)
Issue -
State: open - Opened by richardxoldman 3 months ago
- 1 comment
#917 - Fix chat templates
Pull Request -
State: closed - Opened by danielhanchen 3 months ago
#916 - Fix Chat Templates
Pull Request -
State: closed - Opened by danielhanchen 3 months ago
#915 - Llama 3 template issue
Issue -
State: open - Opened by minipasila 3 months ago
- 4 comments
#914 - PermissionError: [WinError 5]
Issue -
State: open - Opened by silentgameshub 3 months ago
- 1 comment
#913 - Converting unsloth finetuned model to AWQ using autoawq package.
Issue -
State: open - Opened by fusesid 3 months ago
- 8 comments
#912 - Why do the number of parameters of unsloth models differ from the original model?
Issue -
State: open - Opened by Keramatfar 3 months ago
- 3 comments
#911 - Tinyllama issues
Issue -
State: open - Opened by Srini-98 3 months ago
- 3 comments
Labels: currently fixing
#911 - Tinyllama issues
Issue -
State: closed - Opened by Srini-98 3 months ago
- 4 comments
Labels: currently fixing
#910 - Providing more flexibility for users to customize their llama when using LoRA
Pull Request -
State: closed - Opened by Brownwang0426 3 months ago
- 1 comment
#909 - unsloth's FastLanguageModel target_modules does not support custom attention layer
Issue -
State: closed - Opened by Brownwang0426 3 months ago
- 2 comments
#909 - unsloth's FastLanguageModel target_modules does not support custom attention layer
Issue -
State: closed - Opened by Brownwang0426 3 months ago
- 2 comments
#908 - Request for Support: Phi-3 Vision Model
Issue -
State: open - Opened by rahatarinasir 3 months ago
- 1 comment
Labels: feature request
#908 - Request for Support: Phi-3 Vision Model
Issue -
State: open - Opened by rahatarinasir 3 months ago
- 2 comments
Labels: feature request
#907 - google/recurrentgemma-9b-it support
Issue -
State: closed - Opened by ammary25 3 months ago
- 1 comment
Labels: feature request
#906 - Fix DPO stats
Pull Request -
State: closed - Opened by danielhanchen 3 months ago
#905 - Unsloth should probably not alter or delete source models.
Issue -
State: open - Opened by rwl4 3 months ago
- 1 comment
#904 - The issue regarding the parameter settings during llama3.1 training.
Issue -
State: closed - Opened by Xtian-hub 3 months ago
- 3 comments
#903 - Why attention_mask is ignored?
Issue -
State: closed - Opened by Theodotus1243 3 months ago
- 1 comment
#902 - Torch 2.4, Xformers>0.0.27, TRL>0.9, Python 3.12 + bug fixes
Pull Request -
State: closed - Opened by danielhanchen 3 months ago
#901 - Add ability to set flash_attn_func(deterministic=True)
Issue -
State: closed - Opened by Theodotus1243 3 months ago
- 1 comment
#900 - Create mode.bin file when we are saving the model.
Issue -
State: open - Opened by JaynouOliver 3 months ago
- 2 comments
#899 - I ran into errors when running the conda installation locally on my Dell Precision 7920 with the RTX 3090
Issue -
State: open - Opened by kirahman2 3 months ago
- 1 comment
#898 - Not able to push the whole model to hugging face after training : help urgent , Hackathon submission EOD
Issue -
State: open - Opened by Data-Scientist-Sahil 3 months ago
- 6 comments
#897 - Feature Request: Loading the model again requires GPU even though quantization has been done.
Issue -
State: closed - Opened by ahmedembeddedxx 3 months ago
- 1 comment
#896 - Load Unsloth-FT-Merged-Model with AutoModel Attribute Error
Issue -
State: open - Opened by carstendraschner 3 months ago
- 6 comments
#895 - Error: one of the variables needed for gradient computation has been modified by an inplace operation
Issue -
State: open - Opened by YZHang2333 3 months ago
- 2 comments
Labels: currently fixing
#894 - Does unsloth support token classification using phi3?
Issue -
State: open - Opened by xinyudong93 3 months ago
- 1 comment
#893 - Supporting "LoRA+: Efficient Low Rank Adaptation of Large Models"
Issue -
State: open - Opened by fzyzcjy 3 months ago
- 1 comment
#892 - LGAI-EXAONE/EXAONE-3.0-7.8B-Instruct support
Issue -
State: closed - Opened by Ammar-Alnagar 3 months ago
- 2 comments
Labels: feature request
#891 - Notebook for Fine-Tuning on Chat Dataset and Batch Inference
Issue -
State: open - Opened by binhmed2lab 3 months ago
- 1 comment
#890 - Could you please upgrade the trl library to the latest version?
Issue -
State: open - Opened by ArcherShirou 3 months ago
- 3 comments
Labels: fixed - pending confirmation
#889 - AttributeError: torch._dynamo.config.vocab_size does not exist
Issue -
State: open - Opened by nyl199310 3 months ago
- 8 comments
Labels: fixed - pending confirmation
#888 - Cache only has 0 layers, attempted to access layer with index 0
Issue -
State: open - Opened by arturwplantecs 3 months ago
- 2 comments
#887 - Fix tokenizers
Pull Request -
State: closed - Opened by danielhanchen 3 months ago
#886 - support torch=2.4.0 and python=3.12
Issue -
State: open - Opened by NiuBlibing 3 months ago
- 8 comments
Labels: currently fixing
#885 - Phi-3 models not supported
Issue -
State: closed - Opened by umasehs 3 months ago
- 3 comments
Labels: fixed - pending confirmation
#884 - [Urgent] Llama3 NOT Working in PPO Trainer
Issue -
State: open - Opened by yuan-xia 3 months ago
- 5 comments
#883 - Extra response with fine-tuned model
Issue -
State: open - Opened by LiuAlex1109 3 months ago
- 2 comments
#882 - The fine-tuning of the 2024.8 version has become very poor compared to the previous version.
Issue -
State: open - Opened by githubzuoyi 3 months ago
- 12 comments
Labels: currently fixing
#881 - Dubious !
Issue -
State: open - Opened by dromeuf 3 months ago
- 23 comments
Labels: currently fixing, URGENT BUG
#880 - Outdated tokenizer chat template for llama3.1 8B (`unsloth/Meta-Llama-3.1-8B-Instruct-bnb-4bit`)
Issue -
State: closed - Opened by ivsanro1 3 months ago
- 5 comments
Labels: fixed - pending confirmation
#879 - xFormers can't load C++/CUDA extensions
Issue -
State: closed - Opened by DaddyCodesAlot 3 months ago
- 2 comments
#878 - how to train on non instruction dataset .
Issue -
State: open - Opened by hemangjoshi37a 3 months ago
- 8 comments
#877 - Issue Report: Inconsistent Behavior and Meaningless Output
Issue -
State: closed - Opened by seolhokim 3 months ago
- 4 comments
#876 - Request: Flux (Diffusion transformer)
Issue -
State: open - Opened by RefractAI 3 months ago
- 3 comments
Labels: currently fixing, feature request
#876 - Request: Flux (Diffusion transformer)
Issue -
State: open - Opened by RefractAI 3 months ago
- 3 comments
Labels: currently fixing, feature request