Text Generation
PEFT
Safetensors
GGUF
English
Thai
lora
qwen3.5-moe
qwen3.6
reasoning
kimi-k2.6
claude-opus
distillation
weight-diff
svd
Instructions to use hotdogs/qwen3.6-35b-opus-to-kimi-lora with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- PEFT
How to use hotdogs/qwen3.6-35b-opus-to-kimi-lora with PEFT:
from peft import PeftModel from transformers import AutoModelForCausalLM base_model = AutoModelForCausalLM.from_pretrained("lordx64/Qwen3.6-35B-A3B-Claude-4.7-Opus-Reasoning-Distilled") model = PeftModel.from_pretrained(base_model, "hotdogs/qwen3.6-35b-opus-to-kimi-lora") - Notebooks
- Google Colab
- Kaggle
| @article{uka2026weightdiff, | |
| title = {Weight-Diff SVD Extraction: Zero-Shot LoRA Adapter Synthesis from Full-Model Deltas}, | |
| author = {UKA and hotdogs}, | |
| year = {2026}, | |
| month = may, | |
| note = {Hermes Agent, Nous Research}, | |
| url = {https://huggingface.co/hotdogs/qwen3.6-35b-opus-to-kimi-lora}, | |
| abstract = {We present a novel technique for extracting Low-Rank Adaptation (LoRA) adapters directly from the weight difference between two fine-tuned models sharing a common base, without any training. By computing the element-wise delta between model weights and applying truncated Singular Value Decomposition (SVD) tensor-by-tensor, we compress a 70 GB full-model difference into a 7.2 MB rank-16 LoRA adapter. We demonstrate this on Qwen3.6-35B-A3B, extracting the reasoning-style delta between Claude Opus 4.7 and Kimi K2.6 distilled variants. The resulting adapter converts Opus-style concise reasoning (mean 849 tokens) into Kimi-style deliberate reasoning (mean 2,933 tokens) while requiring only 3 GB of VRAM and 3 minutes of CPU compute.} | |
| } | |
| @software{uka2026extract, | |
| title = {extract\_lora\_diff.py: Weight-Diff SVD LoRA Extraction Script}, | |
| author = {UKA}, | |
| year = {2026}, | |
| url = {https://huggingface.co/hotdogs/qwen3.6-35b-opus-to-kimi-lora/blob/main/extract_lora_diff.py} | |
| } | |