nabin2004/AOS-qwen25-coder-7b-manim-sft

AOS Qwen2.5-Coder-7B Manim SFT (LoRA)

LoRA adapter fine-tuned on Code Agent trajectories for Manim animation generation via multi-turn tool calling (run_code + workspace tools).

Model URL: https://huggingface.co/nabin2004/AOS-qwen25-coder-7b-manim-sft

Base model

This adapter is trained on top of Qwen/Qwen2.5-Coder-7B-Instruct (Apache 2.0). See docs/MODEL_SELECTION.md for the selection rationale.

Training data

Fine-tuned on nabin2004/AOS-Trajectories using the AOS Phase 1 SFT trainer (apps/sft).

Usage

Load with PEFT (Python)

import os
import torch
from peft import PeftModel
from transformers import AutoModelForCausalLM, AutoTokenizer, BitsAndBytesConfig

token = os.environ.get("HF_TOKEN")
base_id = "Qwen/Qwen2.5-Coder-7B-Instruct"
adapter_id = "nabin2004/AOS-qwen25-coder-7b-manim-sft"

bnb = BitsAndBytesConfig(
    load_in_4bit=True,
    bnb_4bit_use_double_quant=True,
    bnb_4bit_quant_type="nf4",
    bnb_4bit_compute_dtype=torch.bfloat16,
)

base = AutoModelForCausalLM.from_pretrained(
    base_id,
    quantization_config=bnb,
    device_map="auto",
    torch_dtype=torch.bfloat16,
    token=token,
)
model = PeftModel.from_pretrained(base, adapter_id, token=token)
tokenizer = AutoTokenizer.from_pretrained(adapter_id, token=token)
model.eval()

Inference with AOS tool loop

The adapter is trained on multi-turn Code Agent tool calls, not single-turn prose. Use the AOS inference script:

cd apps/sft
uv run python infer.py \
  --adapter-dir nabin2004/AOS-qwen25-coder-7b-manim-sft \
  --prompt "Create a short Manim scene explaining eigenvectors in 2D."