gemma-2-9b

Instruction Residuals

This repository contains instruction residuals (delta weights) computed as the parameter-wise difference between google/gemma-2-9b-it and google/gemma-2-9b.

Apply these residuals to the base model to reconstruct the instruction-tuned weights without retraining.

Usage

from transformers import AutoModelForCausalLM, AutoTokenizer
from residuals import Residuals

base = AutoModelForCausalLM.from_pretrained("google/gemma-2-9b")
tok = AutoTokenizer.from_pretrained("google/gemma-2-9b")

res = Residuals.from_pretrained("residuals/gemma-2-9b")
res.apply(base, base_tokenizer=tok)

Provenance

• Created at: 2025-10-25T18:20:46.383662+00:00
• DType: float32
• Parameters: 465
• Shapes hash: cae28f7487d2420167b7006d3b12ce510ef0f7283d703cfcd3bd570430900be2
• Names hash: f7de592aab9eee1f49ae785aecd6e6e338ff6b5e2a10ea809381e275141fe06e
• Base model: google/gemma-2-9b
• Instruction model: google/gemma-2-9b-it

Files

• model.safetensors: Serialized residual tensors (safetensors format).
• (optional) model.safetensors.index.json + shard files model-00001-of-000N.safetensors, ... for multi-part weights.
• config.json: Residuals metadata and provenance.
• tokenizer files: Saved tokenizer for compatibility.

About this format

These are additive residuals (task vectors). Applying them to the base model's parameters reconstructs the instruction-tuned model.

Tools

Generated with the residuals Python package. Install via: pip install residuals.

• PyPI: https://pypi.org/project/residuals/
• Source: https://github.com/omarish/residuals