metadata
library_name: residuals
base_model: ibm-granite/granite-4.0-h-tiny-base
base_model_relation: adapter
instruct_model: ibm-granite/granite-4.0-h-tiny
pipeline_tag: text-generation
tags:
- residuals
- delta
- task-arithmetic
- finetune
Instruction Residuals
This repository contains instruction residuals (delta weights) computed as the parameter-wise difference between ibm-granite/granite-4.0-h-tiny and ibm-granite/granite-4.0-h-tiny-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-tiny-base")
tok = AutoTokenizer.from_pretrained("ibm-granite/granite-4.0-h-tiny-base")
res = Residuals.from_pretrained("residuals/granite-4.0-h-tiny")
res.apply(base, base_tokenizer=tok)
Provenance
- Created at: 2025-10-25T17:44:55.705169+00:00
- DType: float32
- Parameters: 587
- Shapes hash: 6dc91196b3c1b84ed135c5dcd5152fd5f2dcf49c5865e1cd097b1c26693cab11
- Names hash: c8c3f98304c698c0e6fad6202d193586a30c58e61d250b2972183fa48fe29a9b
- Base model:
ibm-granite/granite-4.0-h-tiny-base - Instruction model:
ibm-granite/granite-4.0-h-tiny
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.