residuals/granite-4.0-h-micro

Instruction Residuals

This repository contains instruction residuals (delta weights) computed as the parameter-wise difference between ibm-granite/granite-4.0-h-micro and ibm-granite/granite-4.0-h-micro-base.

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("ibm-granite/granite-4.0-h-micro-base")
tok = AutoTokenizer.from_pretrained("ibm-granite/granite-4.0-h-micro-base")

res = Residuals.from_pretrained("residuals/granite-4.0-h-micro")
res.apply(base, base_tokenizer=tok)

Provenance

• Created at: 2025-10-25T17:40:59.623585+00:00
• DType: float32
• Parameters: 467
• Shapes hash: 910db9fc5770fda73a85ced6cea0e6e2a053e0346b9eac50091b0dae3023ad59
• Names hash: 82d0aee30bf5d9833ffe7352a9e912760015befabf4dddd308608cbc395977ec
• Base model: ibm-granite/granite-4.0-h-micro-base
• Instruction model: ibm-granite/granite-4.0-h-micro

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