threadweaver-qwen3-8b-sft

This repository contains the final exported model from the task_sft run in expts/02_tw_repro.

Source checkpoint

• Local final checkpoint: task_sft/output/checkpoint/checkpoint-60
• Final checkpoint timestamp: 2026-04-10 04:32 UTC
• Base model: Qwen/Qwen3-8B
• Architecture: Qwen3ForCausalLM
• Max position embeddings: 40960

Files

This upload contains the inference-ready exported model weights, tokenizer files, and generation config from the final checkpoint.

Quick start

from transformers import AutoModelForCausalLM, AutoTokenizer

model_id = "parallel-reasoner/threadweaver-qwen3-8b-sft"
tokenizer = AutoTokenizer.from_pretrained(model_id)
model = AutoModelForCausalLM.from_pretrained(model_id, torch_dtype="auto", device_map="auto")