Downloads · 30 days
3
5% of all-time downloads
satoyutaka/LLM2026_main_v13_dpo
LLM2026_main_v13_dpo 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. The card lists the license as apache-2.0.
V13 "Silent Brain" is the flagship merged model for the LLM2026 competition, designed to achieve the ultimate balance between Extreme Intelligence and Absolute Silence.
Downloads · 30 days
3
5% of all-time downloads
All-time downloads
65
Public
Parameters
4B
8.1 GB on disk
Likes
0
Public
Click a slice to open those files.
.safetensors8 GB · 100%
From the Hugging Face model README
V13 "Silent Brain" is the flagship merged model for the LLM2026 competition, designed to achieve the ultimate balance between Extreme Intelligence and Absolute Silence.
This model builds upon the structural precision of the 0.75 SFT base (satoyutaka/LLM2026_SFT_0_again) and applies a specialized "Hidden CoT" DPO alignment. Unlike previous iterations, V13 internalizes its reasoning process, providing highly accurate structured data without any conversational overhead or explicit Chain-of-Thought in the output.
V13 follows a rigorous sequential alignment and merging process:
Qwen/Qwen3-4B-Instruct-2507satoyutaka/LLM2026_SFT_0_again), which was trained on 4,000 surgical data samples to establish fundamental structural discipline.The final alignment stage (V13 DPO) uses a novel "Hidden Chain-of-Thought" approach:
q_proj, k_proj, v_proj, o_proj, gate_proj, up_proj, down_proj) to maximize reasoning capacity.This model strictly adheres to competition ethical guidelines:
u-10bei/dpo-dataset-qwen-cot and structured-hard-sft-4k.jsonl.v13_data_generation.py.This is a FULLY MERGED model. You can load it directly without applying additional adapters:
from transformers import AutoModelForCausalLM, AutoTokenizer
model_id = "satoyutaka/LLM2026_main_v13_dpo"
tokenizer = AutoTokenizer.from_pretrained(model_id)
model = AutoModelForCausalLM.from_pretrained(model_id, torch_dtype="auto", device_map="auto")
In V13, we completely revamped the chat template to match the exact format used during the DPO training phase (### Instruction: and ### Response:).
Why we did this:
By default, the Qwen3 base model utilizes a standard conversational template (<|im_start|>user\n...<|im_end|>). If this conversational template is used during inference, it subtly cues the model to respond in a conversational, "helpful" manner (i.e., chattiness, explaining its reasoning, or adding "Here is the JSON...").
Since the V13 DPO training strictly punished this chattiness by setting the "chatty CoT output" as the rejected response and the "pure structured data" as the chosen response under the ### Instruction: format, we discovered that the model's "Silent Brain" is only reliably triggered when the inference template exactly matches the training template.
Therefore, the tokenizer_config.json was manually updated so apply_chat_template automatically formats the prompt into the trigger format:
"chat_template": "{%- for message in messages %}{%- if message['role'] == 'system' and message['content'] %}{{ message['content'] + '\\n\\n' }}{%- elif message['role'] == 'user' %}### Instruction:\\n{{ message['content'] }}\\n\\n### Response:\\n{%- elif message['role'] == 'assistant' %}{{ message['content'] }}{%- endif %}{%- endfor %}"
Note: This template override ensures that lists ([...]), standard objects ({...}), and formats like CSV/YAML can be generated dynamically without being forced by a prefill hack (like appending { to the generation prompt).
exp_20260228_v13_silent_dpoV13 "Silent Brain (沈黙する知能)" は、LLM2026コンペティションにおいて 「極致の知能」 と 「絶対的な静粛性」 を両立させるために開発された最終マージモデルです。
本モデルは、4,000件の精密データで学習された 0.75 SFTベース (satoyutaka/LLM2026_SFT_0_again) の堅牢な構造化能力を土台とし、そこに「Hidden CoT (内面化された推論)」DPOアライメントを施しています。従来のモデルと異なり、思考プロセスを内部的に完結(内面化)させることで、出力には一切の推論過程や前置きを出さず、極めて高い精度で構造化データのみを生成します。
V13は、以下の厳格なステップを経て構築されました。
Qwen/Qwen3-4B-Instruct-2507最終段階のアライメントでは、思考を「出力」ではなく「重み」に焼き付ける手法を採用しました。
q_proj, k_proj, v_proj, o_proj, gate_proj, up_proj, down_proj) を学習対象。本モデルは、データの倫理性に関するコンペティション規約を厳格に遵守しています。
u-10bei/dpo-dataset-qwen-cot および structured-hard-sft-4k.jsonl をベースに加工。v13_data_generation.py を用いて、250セットの高品質なDPOペアを生成。本モデルは完全にマージ済みです。アダプタの適用なしでそのままロード可能です。
exp_20260228_v13_silent_dpo