---
base_model: Qwen/Qwen3-4B-Instruct-2507
datasets:
  - Gen-Verse/Open-AgentRL-SFT-3K
language:
  - en
license: apache-2.0
library_name: transformers
pipeline_tag: text-generation
tags:
  - agent
  - tool-use
  - sft
  - multi-turn
  - code-interpreter
  - open-agentrl
---

# Qwen3-4B-Agent-SFT-True

This repository contains a **full fine-tuned model** (not LoRA adapter) based on
**Qwen3-4B-Instruct-2507**, trained with multi-turn agentic SFT using the
[Open-AgentRL](https://github.com/Gen-Verse/Open-AgentRL) framework (verl FSDP SFT Trainer).

## Training Configuration

| Parameter | Value |
|---|---|
| Base model | `Qwen/Qwen3-4B-Instruct-2507` |
| Method | Full fine-tuning (FSDP, bfloat16) |
| Max sequence length | 32,768 |
| Epochs | 10 |
| Train batch size | 16 |
| Micro batch size per GPU | 1 |
| Truncation | right |
| Trainer | `verl.trainer.fsdp_sft_trainer` |

## Dataset

- **Name**: [Gen-Verse/Open-AgentRL-SFT-3K](https://huggingface.co/datasets/Gen-Verse/Open-AgentRL-SFT-3K)
- **Samples**: 3,000 multi-turn conversations
- **Source**: Original Open-AgentRL SFT dataset (real End-to-End agentic trajectories)

## Usage

```python
from transformers import AutoModelForCausalLM, AutoTokenizer
import torch

model_id = "y-ohtani/qwen3-4b-agent-sft-true"

tokenizer = AutoTokenizer.from_pretrained(model_id)
model = AutoModelForCausalLM.from_pretrained(
    model_id,
    torch_dtype=torch.bfloat16,
    device_map="auto",
)

messages = [
    {"role": "user", "content": "Solve the equation x^2 - 5x + 6 = 0 step by step."}
]
text = tokenizer.apply_chat_template(messages, tokenize=False, add_generation_prompt=True)
inputs = tokenizer(text, return_tensors="pt").to(model.device)
outputs = model.generate(**inputs, max_new_tokens=2048)
print(tokenizer.decode(outputs[0], skip_special_tokens=True))
```

## Sources & Terms

| Component | Source | License |
|---|---|---|
| Base model | [Qwen/Qwen3-4B-Instruct-2507](https://huggingface.co/Qwen/Qwen3-4B-Instruct-2507) | Apache-2.0 |
| SFT dataset | [Gen-Verse/Open-AgentRL-SFT-3K](https://huggingface.co/datasets/Gen-Verse/Open-AgentRL-SFT-3K) | -- |
| Training framework | [Open-AgentRL](https://github.com/Gen-Verse/Open-AgentRL) (verl) | Apache-2.0 |

Users must comply with the base model license and dataset terms.