Hugging Face's logo

codellama
/
CodeLlama-7b-Python-hf

Text Generation
Transformers
PyTorch
Safetensors
code
llama
llama-2
Inference Endpoints
text-generation-inference
Model card Files Community
13

6 shard model checkoint size

#4
by bzz2 - opened
Discussion
bzz2

Probably something went wrong during the conversion as 6-shard checkpoint is ~28Gb instead of 14Gb for the 2-shard version.

osanseviero
 Code Llama org

Thanks! We'll fix this. We pushed the weights in float32 rather than bfloat16, which likely explained the discrepancy

bzz2
•
edited Aug 28, 2023

Thank you for the update! Closing, as it has been addressed by #8

bzz2 changed discussion status to closed

Sign up or log in to comment