lxe/simple-llm-finetuner

Issues

• How can I use the finetuned model with text-generation-webui or KoboldAI?

  #16 opened 3 months ago by IIIIIIIllllllllIIIII
  4

• Slow generation speed: around 10 minutes / loading forever on rtx3090 with 64gb ram....

  #17 opened 3 months ago by IIIIIIIllllllllIIIII
  3

• How do I merge trained Lora an Llama7b weight?

  #41 opened 3 months ago by IIIIIIIllllllllIIIII
  2

• How to work with vLLM such as LLAVA

  #63 opened 10 months ago by dinhquangsonimip
  0

• How should I prepare the dataset for generative question answering on the private documents?

  #38 opened 2 years ago by AayushSameerShah
  50

• Question: Native windows support

  #23 opened a year ago by Paillat-dev
  3

• will this work with quantized GGUF files?

  #62 opened a year ago by sengiv
  0

• [Request] Retrain adapter from checkpoint?

  #61 opened a year ago by zaanind
  0

• Performance after FineTuning

  #47 opened 2 years ago by Datta0
  3

• About llama-2-70B fine-tuning

  #60 opened 2 years ago by RickMeow
  0

• Getting the repo id error from the web interface

  #59 opened 2 years ago by matt7salomon
  0

• How to use CPU instead of GPU

  #33 opened 2 years ago by Shreyas-ITB
  2

• AMD GPU compability or CPU

  #58 opened 2 years ago by LeLaboDuGame
  0

• RuntimeError: expected scalar type Half but found Float

  #52 opened 2 years ago by jasperan
  3

• M1/M2 Metal support?

  #57 opened 2 years ago by itsPreto
  0

• [Request] Mac ARM support

  #56 opened 2 years ago by voidcenter
  0

• RuntimeError: unscale_() has already been called on this optimizer since the last update().

  #54 opened 2 years ago by junxu-ai
  3

• [Request] QLoRA support

  #55 opened 2 years ago by CoolOppo
  0

• In trainer.py, ignore the last token is not suitable for all situations.

  #53 opened 2 years ago by HCTsai
  0

• `LLaMATokenizer` vs `LlamaTokenizer` class names

  #22 opened 2 years ago by vadi2
  5

• Question: Is fine tuning suitable for factual answers from custom data, or is it better to use vector databases and use only the relevant chunk in the prompt for factual answers?

  #4 opened 2 years ago by petrbrzek
  2

• Multi GPU running

  #51 opened 2 years ago by Shashika007
  0

• Issue in train in colab

  #42 opened 2 years ago by fermions75
  7

• Getting OOM

  #46 opened 2 years ago by alior101
  2

• Error: Adapter lora/decapoda-research_llama-{ADAPTER_NAME} not found.

  #44 opened 2 years ago by 64-bit
  0

• Suggestion to improve UX

  #32 opened 2 years ago by ch3rn0v
  3

• "The tokenizer class you load from this checkpoint is 'LLaMATokenizer'."

  #40 opened 2 years ago by IIIIIIIllllllllIIIII
  3

• Verbose function to find out what leads to crash during training?

  #39 opened 2 years ago by IIIIIIIllllllllIIIII
  1

• question: could the model trained be used for alpaca.cpp?

  #20 opened 2 years ago by goog
  2

• AttributeError: type object 'Dataset' has no attribute 'from_list'

  #36 opened 2 years ago by Datta0
  3

• "error" in training - AttributeError: 'CastOutputToFloat' object has no attribute 'weight', RuntimeError: Only Tensors of floating point and complex dtype can require gradients

  #29 opened 2 years ago by GreenTeaBD
  5

• Training using long stories instead of question/response

  #24 opened 2 years ago by leszekhanusz
  3

• how to finetune with 'system information'

  #30 opened 2 years ago by mhyeonsoo
  1

• Attempting to use 13B in the simple tuner -

  #28 opened 2 years ago by Atlas3DSS
  2

• Not a problem - but like people should know

  #26 opened 2 years ago by Atlas3DSS
  2

• Error during Training RuntimeError: mat1 and mat2 shapes cannot be multiplied (511x2 and 3x4096)

  #25 opened 2 years ago by kasakh
  2

• Traceback during inference.

  #6 opened 2 years ago by Hello1024
  8

• Examples to get started with

  #11 opened 2 years ago by vadi2
  4

• How the finetuning output looks like?

  #27 opened 2 years ago by mhyeonsoo
  1

• Can Nivdia 3090 with 24G video memory support finetune?

  #7 opened 2 years ago by pczzy
  4

• Inference doesn't work after training

  #10 opened 2 years ago by vadi2
  2

• Finetuning in unsupported language

  #15 opened 2 years ago by jumasheff
  2

• Host on Hugging Face Spaces

  #19 opened 2 years ago by osanseviero
  4

• Inference output text keeps running on...

  #1 opened 2 years ago by lxe
  1

• Inference works just once

  #12 opened 2 years ago by vadi2
  12

• (WSL2) - No GPU / Cuda detected....

  #13 opened 2 years ago by IIIIIIIllllllllIIIII
  6

• Collecting info on memory requirements

  #2 opened 2 years ago by jmiskovic
  1

• Is CUDA 12.0 supported?

  #8 opened 2 years ago by vadi2
  1

• Where are the downloaded ".bin" files for the llama model stored on the disk?

  #3 opened 2 years ago by ashishb
  2