4 Pull requests merged by 2 people

• added readme to broswer use usecase

  #879 merged Feb 10, 2025

• update root README with absolute path to fix https://pypi.org/project/llama-cookbook/ broken links

  #880 merged Feb 10, 2025

• ignore medium and linkedin

  #882 merged Feb 10, 2025

• added readme for customerservice chatbots

  #881 merged Feb 10, 2025

1 Pull request opened by 1 person

• remove key

  #883 opened Feb 12, 2025

17 Issues closed by 2 people

• Fine-tuning with LLaMA_Recipes does not retain dataset-specific information

  #800 closed Feb 11, 2025

• I couldn't find the code about Video Encoder in llama 3.2 vision

  #795 closed Feb 11, 2025

• finetuning using notebook on custom dataset

  #788 closed Feb 11, 2025

• The notebook titled "Fine-tuning with Multi GPU" does not seem to generate the dataset correctly. Possible get_dataloader bug.

  #777 closed Feb 11, 2025

• TemplateError: Prompting with images is incompatible with system messages.

  #774 closed Feb 11, 2025

• Adding new vocab doesn't saved the model

  #773 closed Feb 11, 2025

• RuntimeError: Expected all tensors to be on the same device, but found at least two devices, cuda:0 and cpu! error for chatcompletion.py with llama 3.2 instruct model

  #771 closed Feb 11, 2025

• Found two forward recomputation exist in a single backward when using FSDP with activation checkpointing

  #740 closed Feb 11, 2025

• Learning rate scheduler

  #738 closed Feb 11, 2025

• llama3.2 fine tuning generates repeated pattern towards the end of one epoch

  #735 closed Feb 11, 2025

• Checkpoint feature via steps instead of epoch

  #724 closed Feb 11, 2025

• Configs like `train_config` should be named using PascalCase

  #722 closed Feb 11, 2025

• Recipe for fine-tuning with lower resolution images

  #714 closed Feb 11, 2025

• Merging lora adapter with Llama 3.2 vision

  #702 closed Feb 11, 2025

• llama finetune.py throws pytorch tensor datatype error with 4 bit quantization

  #675 closed Feb 11, 2025

• The EOS and BOS token setting when contine pretraining Llama3.1

  #648 closed Feb 11, 2025

• Speaker Consistency Issue in TTS Workflow (NotebookLlama)

  #870 closed Feb 11, 2025

1 Issue opened by 1 person

• safetensors_rust.SafetensorError: Error while deserializing header: HeaderTooLarge

  #884 opened Feb 13, 2025

4 Unresolved conversations

Sometimes conversations happen on old items that aren’t yet closed. Here is a list of all the Issues and Pull Requests with unresolved conversations.

• Finetuning llama 3.1 7B models using FSDP+4-bit-quantization+PEFT not reducing GPU consumption

  #872 commented on Feb 9, 2025 • 0 new comments

• Existing Llama Guard recipes uses "O" rather than "S" as category code prefix.

  #849 commented on Feb 12, 2025 • 0 new comments

• continue pre-training Example

  #820 commented on Feb 12, 2025 • 0 new comments

• deprecate OctoAI

  #873 commented on Feb 12, 2025 • 0 new comments