Downloads · 30 days
0
winglian/llama-3-gradientai-1m-split
llama-3-gradientai-1m-split is a machine learning model from winglian. Use it for the machine learning task on the model card, and read the license before you ship it in a product.
LoRA adapter based on GradientAI's 1M context Llama-3 8B Instruct finetune. I found that rank 1024 is not sufficient to capture the delta weights in the qproj and oproj, so I've created seperate adapters for those mod…
Downloads · 30 days
0
Access
Public
Updated Apr 30, 2024
Repo size
9 GB
Likes
1
Public
Click a slice to open those files.
.safetensors9 GB · 100%
From the Hugging Face model README
LoRA adapter based on GradientAI's 1M context Llama-3 8B Instruct finetune. I found that rank 1024 is not sufficient to capture the delta weights in the q_proj and o_proj, so I've created seperate adapters for those modules vs the k-v projection modules.