Accio-Lab

Metis-8B-RL

README

License: apache-2.0

Highlights

98% → 2% Tool Calls — Reduces blind tool invocation by orders of magnitude.
SOTA Performance — Best accuracy across 13 benchmarks among open-source 8B agentic models.
Meta-Cognitive Wisdom — Learns when to use tools, not just how.

Model Details

Table with columns: Attribute, Value
Attribute	Value
Base model	Qwen3-VL-8B-Instruct
SFT checkpoint	Metis-8B-ColdStart
RL algorithm	HDPO (Hierarchical Decoupled Policy Optimization)
Training data	Metis-RL (~5K prompts)
License	Apache-2.0

HDPO Training Hyperparameters

Table with columns: Hyperparameter, Value
Hyperparameter	Value
Batch size	128
Rollouts per prompt (G)	16
Learning rate	1e-6
KL coefficient	0
Loss weights	w_acc = 1.0, w_tool = 0.15
Max response length	16,384 tokens

Method: Hierarchical Decoupled Policy Optimization (HDPO)

Current agentic multimodal models suffer from blind tool invocation — they reflexively call external tools even when queries are directly resolvable from the visual context. Existing RL methods attempt to fix this by coupling accuracy and tool-efficiency into a single scalar reward, but this creates an irreconcilable optimization dilemma.

HDPO resolves this through three key components:

Dual Reward Design — An accuracy reward (r_acc) and a tool-efficiency reward (r_tool) that is conditioned on correctness.
Decoupled Advantage Estimation — Accuracy advantages are computed over all rollouts; tool efficiency advantages are computed exclusively over correct rollouts (conditional GRPO).
Hierarchical Policy Update — Two independent clipped surrogate losses combined as L_HDPO = w_acc · L_GRPO(A_acc) + w_tool · L_GRPO(A_tool).

This naturally induces an implicit curriculum: first learn to be correct, then learn to be efficient.

Evaluation Results

Perception and Document Understanding

Table with columns: Model, V*Bench, HR4K, HR8K, TreeBench, MME-RW, SEED2+, CharXiv(DQ), CharXiv(RQ)
Model	V*Bench	HR4K	HR8K	TreeBench	MME-RW	SEED2+	CharXiv(DQ)	CharXiv(RQ)
Qwen3-VL-8B-Instruct	86.4	78.9	74.6	40.7	61.9	71.0	83.0	46.3
DeepEyesV2	81.8	77.9	73.8	42.5	64.9	70.5	78.6	48.9
SenseNova-MARS-8B	92.2	83.1	78.4	-	67.9	-	-	-
Skywork-R1V4-30B-A3B	88.0	82.8	79.8	-	71.4	-	-	-
Metis (Ours)	91.1	83.5	82.0	45.2	70.3	72.5	83.4	54.1

Mathematical and Logical Reasoning

Table with columns: Model, MathVista, MathVerse, WeMath, DynaMath, LogicVista, Avg.
Model	MathVista	MathVerse	WeMath	DynaMath	LogicVista	Avg.
Qwen3-VL-8B-Instruct	76.3	61.3	38.8	65.5	54.9	59.4
DeepEyesV2	71.9	52.7	38.1	57.2	48.7	53.7
Metis (Ours)	78.0	65.9	65.2	69.2	56.2	66.9

Usage

Please refer to the GitHub repository for full installation and inference instructions.

Installation

bash

git clone https://github.com/Accio-Lab/Metis.gitcd Metispip install -e verlpip install -e ".[vllm,search_tool,python_code_dep]"

Citation

bibtex

@article{yan2026metis,  title={Act Wisely: Cultivating Meta-Cognitive Tool Use in Agentic Multimodal Models},  author={Yan, Shilin and Tong, Jintao and Xue, Hongwei and Tang, Xiaojun and Wang, Yangyang and Shi, Kunyu and Zhang, Guannan and Li, Ruixuan and Zou, Yixiong},  journal={arXiv preprint arXiv:2604.08545},  year={2026}}

Acknowledgments

Metis is built upon verl, verl-tool, and Qwen3-VL.

Available on FriendliAI

Dedicated Endpoints

Run this model inference on single tenant GPU with unmatched speed and reliability at scale.

Learn more

Model Details

Model Provider

Accio-Lab

Model Tree

Base

Accio-Lab/Metis-8B-ColdStart

Fine-tuned

this model

Input Modalities

TextImage

Output Modalities

Text

Supported Functionality

Dedicated Endpoints

Explore FriendliAI today

Get started Talk to an engineer

README

License: apache-2.0

Highlights

98% → 2% Tool Calls — Reduces blind tool invocation by orders of magnitude.
SOTA Performance — Best accuracy across 13 benchmarks among open-source 8B agentic models.
Meta-Cognitive Wisdom — Learns when to use tools, not just how.

Model Details

Table with columns: Attribute, Value
Attribute	Value
Base model	Qwen3-VL-8B-Instruct
SFT checkpoint	Metis-8B-ColdStart
RL algorithm	HDPO (Hierarchical Decoupled Policy Optimization)
Training data	Metis-RL (~5K prompts)
License	Apache-2.0

HDPO Training Hyperparameters

Table with columns: Hyperparameter, Value
Hyperparameter	Value
Batch size	128
Rollouts per prompt (G)	16
Learning rate	1e-6
KL coefficient	0
Loss weights	w_acc = 1.0, w_tool = 0.15
Max response length	16,384 tokens

Method: Hierarchical Decoupled Policy Optimization (HDPO)

HDPO resolves this through three key components:

Dual Reward Design — An accuracy reward (r_acc) and a tool-efficiency reward (r_tool) that is conditioned on correctness.
Decoupled Advantage Estimation — Accuracy advantages are computed over all rollouts; tool efficiency advantages are computed exclusively over correct rollouts (conditional GRPO).
Hierarchical Policy Update — Two independent clipped surrogate losses combined as L_HDPO = w_acc · L_GRPO(A_acc) + w_tool · L_GRPO(A_tool).

This naturally induces an implicit curriculum: first learn to be correct, then learn to be efficient.

Evaluation Results

Perception and Document Understanding

Table with columns: Model, V*Bench, HR4K, HR8K, TreeBench, MME-RW, SEED2+, CharXiv(DQ), CharXiv(RQ)
Model	V*Bench	HR4K	HR8K	TreeBench	MME-RW	SEED2+	CharXiv(DQ)	CharXiv(RQ)
Qwen3-VL-8B-Instruct	86.4	78.9	74.6	40.7	61.9	71.0	83.0	46.3
DeepEyesV2	81.8	77.9	73.8	42.5	64.9	70.5	78.6	48.9
SenseNova-MARS-8B	92.2	83.1	78.4	-	67.9	-	-	-
Skywork-R1V4-30B-A3B	88.0	82.8	79.8	-	71.4	-	-	-
Metis (Ours)	91.1	83.5	82.0	45.2	70.3	72.5	83.4	54.1

Mathematical and Logical Reasoning

Table with columns: Model, MathVista, MathVerse, WeMath, DynaMath, LogicVista, Avg.
Model	MathVista	MathVerse	WeMath	DynaMath	LogicVista	Avg.
Qwen3-VL-8B-Instruct	76.3	61.3	38.8	65.5	54.9	59.4
DeepEyesV2	71.9	52.7	38.1	57.2	48.7	53.7
Metis (Ours)	78.0	65.9	65.2	69.2	56.2	66.9

Usage

Please refer to the GitHub repository for full installation and inference instructions.

Installation

bash

git clone https://github.com/Accio-Lab/Metis.gitcd Metispip install -e verlpip install -e ".[vllm,search_tool,python_code_dep]"

git clone https://github.com/Accio-Lab/Metis.gitcd Metispip install -e verlpip install -e ".[vllm,search_tool,python_code_dep]"

Citation

bibtex

@article{yan2026metis,  title={Act Wisely: Cultivating Meta-Cognitive Tool Use in Agentic Multimodal Models},  author={Yan, Shilin and Tong, Jintao and Xue, Hongwei and Tang, Xiaojun and Wang, Yangyang and Shi, Kunyu and Zhang, Guannan and Li, Ruixuan and Zou, Yixiong},  journal={arXiv preprint arXiv:2604.08545},  year={2026}}

@article{yan2026metis,  title={Act Wisely: Cultivating Meta-Cognitive Tool Use in Agentic Multimodal Models},  author={Yan, Shilin and Tong, Jintao and Xue, Hongwei and Tang, Xiaojun and Wang, Yangyang and Shi, Kunyu and Zhang, Guannan and Li, Ruixuan and Zou, Yixiong},  journal={arXiv preprint arXiv:2604.08545},  year={2026}}

Acknowledgments

Metis is built upon verl, verl-tool, and Qwen3-VL.