Downloads · 30 days
4
14% of all-time downloads
satoyutaka/Qwen3-4B-AgentBench-llm2025_advance_v5_ironguard-fp32
Qwen3-4B-AgentBench-llm2025_advance_v5_ironguard-fp32 is a machine learning model from satoyutaka. Use it for the machine learning task on the model card, and read the license before you ship it in a product.
This is the V5 "Iron Guard" version of the specialized agent model for AgentBench-comp, based on the Qwen3-4B architecture. This version utilizes DPO (Direct Preference Optimization) to enforce extreme discipline, eli…
Downloads · 30 days
4
14% of all-time downloads
All-time downloads
29
Public
Parameters
4B
16.1 GB on disk
Likes
0
Public
Click a slice to open those files.
.safetensors16.1 GB · 100%
From the Hugging Face model README
This is the V5 "Iron Guard" version of the specialized agent model for AgentBench-comp, based on the Qwen3-4B architecture.
This version utilizes DPO (Direct Preference Optimization) to enforce extreme discipline, eliminate unnecessary verbosity, and ensure 100% adherence to task-specific action formats.
After analyzing V4's failure (where the 7B model suffered from "over-thinking" and timeouts), V5 returns to the agile 4B architecture with a primary focus on Discipline.
Thought -> Action -> Observation loop. It ensures that every response ends with a valid Action: or Final Answer: as required by the AgentBench evaluator.max_length=2048 and full Attention gate updates (q, k, v, o_proj) to maintain complex environment states over long trajectories.Unlike previous SFT-only versions, V5 uses DPO to directly penalize "bad behaviors":
beta=0.1 for subtle but firm style alignment.max_length=2048 to capture long trajectory context.target_modules: Full Attention layers (q_proj, k_proj, v_proj, o_proj).from transformers import AutoModelForCausalLM, AutoTokenizer
import torch
model_id = "satoyutaka/Qwen3-4B-AgentBench-llm2025_advance_1st"
tokenizer = AutoTokenizer.from_pretrained(model_id)
model = AutoModelForCausalLM.from_pretrained(
model_id,
torch_dtype=torch.bfloat16,
device_map="auto",
trust_remote_code=True
)
# Optimized for ReAct format: Thought -> Action -> Observation
本モデルは、AgentBench-compにおける最高スコア(4.5〜5.0)獲得を目指して開発された、Qwen3-4Bベースの第5世代(V5)エージェントモデルです。
**DPO(Direct Preference Optimization)**を適用することで、「鉄の規律(Iron Guard)」を備え、無駄な発話(お喋り)を完全に排除し、タスク遂行に特化した出力スタイルを確立しました。
高知能ながら「考えすぎ」によるタイムアウトやフォーマット違反が多発したV4の反省を活かし、V5では軽量な4Bモデルへの回帰と、徹底した規律の定着を最優先しました。
Thought -> Action -> Observation のループを完璧に守ります。AgentBenchの評価システムが要求する Action: や Final Answer: を、迷いなく正確な書式で出力します。max_length=2048 および Attention全層(q, k, v, o_proj)の更新により、多段階の探索が必要なタスクでも環境情報を正確に保持します。従来のSFT(教師あり学習)に加え、DPOを用いることで「何が悪い行動か」を直接モデルに学習させました。
beta=0.1: 推論能力を損なわずにスタイルを鮮鋭化。max_length=2048: 複雑な推論過程を切り捨てずに学習。target_modules: Attention層すべてを対象とし、文脈保持とフォーマット形成の両面を強化。from transformers import AutoModelForCausalLM, AutoTokenizer
import torch
model_id = "satoyutaka/Qwen3-4B-AgentBench-llm2025_advance_1st"
tokenizer = AutoTokenizer.from_pretrained(model_id)
model = AutoModelForCausalLM.from_pretrained(
model_id,
torch_dtype=torch.bfloat16,
device_map="auto",
trust_remote_code=True
)
# ReAct形式(Thought -> Action -> Observation)に最適化されています