Downloads · 30 days
3
5% of all-time downloads
wvnvwn/Mistral-7B-Instruct-v0.3-gsm8k-ta-lam0p4-a0p5-v1
Mistral-7B-Instruct-v0.3-gsm8k-ta-lam0p4-a0p5-v1 is a machine learning model from wvnvwn. Use it for the machine learning task on the model card, and read the license before you ship it in a product. The card lists the license as apache-2.0.
Merged model produced by Task Arithmetic (Ilharco et al., ICLR 2023).
Downloads · 30 days
3
5% of all-time downloads
All-time downloads
60
Public
Parameters
7.2B
14.5 GB on disk
Likes
0
Public
Click a slice to open those files.
.safetensors14.5 GB · 100%
From the Hugging Face model README
Merged model produced by Task Arithmetic (Ilharco et al., ICLR 2023).
mistralai/Mistral-7B-Instruct-v0.3safety = wvnvwn/Mistral-7B-Instruct-v0.3-hhrlhf-v1, alpha = 0.5downstream = wvnvwn/Mistral-7B-Instruct-v0.3-gsm8k-v1, alpha = 0.5ta0.4bfloat16$$\theta_\text{merged} = \theta_\text{base} + \lambda \cdot M\bigl(\alpha \cdot \tau_\text{safety} + (1-\alpha) \cdot \tau_\text{downstream}\bigr),$$
where tau_i = theta_i_sft - theta_base, and M(.) is the identity for TA,
the Trim-Elect-Disjoint operator for TIES, and DARE's drop-and-rescale
preprocessing (composed with TA) for DARE.
Produced by the GTM merge_cli (see commit history of the GTM repo). Full
metadata is shipped as merge_metadata.json alongside the weights.
Research only. This checkpoint is part of a sweep designed to probe the safety / downstream-utility Pareto frontier; individual α values are NOT recommended for deployment without further evaluation.