Resources

fine-tuning with structured data set

#68 opened 3 days ago by

don412

Questions about chat template format (whether to escape newline characters and special characters or not)

#67 opened 6 days ago by

king-crab

Any plans to release pre-trained (base) models?

#66 opened 8 days ago by

Defetya

Add attention_bias to make TGI work

#64 opened 10 days ago by

philschmid

Datasets used for training

#63 opened 16 days ago by

saransh03sharma

fp16 normal weights

#62 opened 17 days ago by

gioaca

GREAT MODEL

#61 opened 18 days ago by

doberst

Sliding window = 2047?

#60 opened 19 days ago by

skyshine102

Recent change on the rstrip property on special tokens

#59 opened 20 days ago by

xxhansh

fix: update tokenizer config to support `add_generation_prompt=True` and clarify content

#57 opened 22 days ago by

lamhieu

Fix auto-casting with SFT sometimes only up-casting keys and not queries

#56 opened 23 days ago by

roborovski

Help with merging LoRA layers back onto Phi3

#55 opened 24 days ago by

SHIMURA0321

Changed instruction/chat template

#54 opened 25 days ago by

ofirzaf

Phi-3 support for GPU training with MPS on Mac

#52 opened about 1 month ago by

chaishirin

System prompts ignored in chat completions

#51 opened about 1 month ago by

joshuaturner

Fine-tuning is not improving the domain knowledge? it is very complicated, could you help?

#50 opened about 1 month ago by

aaditya

the phi3 model not learning from training data using orpo.

#44 opened about 1 month ago by

Imran1

Leaking Training Data

#40 opened about 1 month ago by

qunitindk

Instruct model needs work

#35 opened about 1 month ago by

Nafnlaus