README.md

---
language:
- zh
- en
pipeline_tag: text-generation
library_name: transformers
---
<div align="center">
  <picture>
      <img src="figures/joyai-logo.png" width="30%" alt="JoyAI-LLM Flash">
  </picture>
</div>
<hr>

<div align="center" style="line-height: 1;">
  <a href="https://huggingface.co/jdopensource" target="_blank"><img
alt="Hugging Face"
src="https://img.shields.io/badge/%F0%9F%A4%97%20Hugging%20Face-JD-ffc107?color=ffc107&logoColor=white"/></a>
  <a
href="https://huggingface.co/jdopensource/JoyAI-LLM-Flash/blob/main/LICENSE"><img
alt="License"
src="https://img.shields.io/badge/License-Modified_MIT-f5de53?&color=f5de53"/></a>
</div>

<p align="center">
<b>📰&nbsp;&nbsp;<a
href="https://arxiv.org/pdf/2604.03044">Tech
Report</a>
</p>



## 1. Model Introduction

JoyAI-LLM Flash is a state-of-the-art medium-sized instruct language model with
3 billion activated parameters and 48 billion total parameters. JoyAI-LLM Flash
was pretrained on 20 trillion text tokens using Muon optimizer, followed by
large-scale supervised fine-tuning (SFT), direct preference optimization (DPO),
and reinforcement learning (RL) across diverse environments. JoyAI-LLM Flash
achieves strong performance across frontier knowledge, reasoning, coding tasks
and agentic capabilities.

### Key Features

- Fibration Policy Optimization: Introduces fiber bundle theory into
  reinforcement learning, proposing a novel optimization framework, FiberPO.
This method is specifically designed to handle the challenges of large-scale
and heterogeneous agent training, improving stability and robustness under
complex data distributions. [paper link](https://arxiv.org/abs/2603.08239)
- Training-Inference Collaboration: apply Muon optimizer with dense MTP,
  develop novel optimization techniques to resolve instabilities while scaling
up, delivering 1.3× to 1.7× the throughput of the non-MTP version.
- Agentic Intelligence: designed for tool use, reasoning, and autonomous
  problem-solving.

## 2. Model Summary

|                                             |                          |
| :-----------------------------------------: | :----------------------: |
|              **Architecture**               | Mixture-of-Experts (MoE) |
|            **Total Parameters**             |           48B            |
|          **Activated Parameters**           |            3B            |
| **Number of Layers** (Dense layer included) |            40            |
|         **Number of Dense Layers**          |            1             |
|       **Attention Hidden Dimension**        |           2048           |
|    **MoE Hidden Dimension** (per Expert)    |           768            |
|        **Number of Attention Heads**        |            32            |
|            **Number of Experts**            |           256            |
|       **Selected Experts per Token**        |            8             |
|        **Number of Shared Experts**         |            1             |
|             **Vocabulary Size**             |           129K           |
|             **Context Length**              |           128K           |
|           **Attention Mechanism**           |           MLA            |
|           **Activation Function**           |          SwiGLU          |
|                   </div>                    |                          |


## 3. Evaluation Results

<table>
<thead>
<tr>
<th align="center">Benchmark</th>
<th align="center"><sup>JoyAI-LLM Flash</sup></th>
<th align="center"><sup>Qwen3-30B-A3B-Instuct-2507</sup></th>
<th align="center"><sup>GLM-4.7-Flash<br>(Non-thinking)</sup></th>
</tr>
</thead>
<tbody>


<tr>
<td align="center" colspan=8><strong>Knowledge &amp; Alignment</strong></td>
</tr>
<tr>
<td align="center" style="vertical-align: middle">MMLU</td>
<td align="center" style="vertical-align: middle"><strong>89.50</strong></td>
<td align="center" style="vertical-align: middle">86.87</td>
<td align="center" style="vertical-align: middle">80.53</td>
</tr>
<tr>
<td align="center" style="vertical-align: middle">MMLU-Pro</td>
<td align="center" style="vertical-align: middle"><strong>81.02</strong></td>
<td align="center" style="vertical-align: middle">73.88</td>
<td align="center" style="vertical-align: middle">63.62</td>
</tr>
<tr>
<td align="center" style="vertical-align: middle">CMMLU</td>
<td align="center" style="vertical-align: middle"><strong>87.03</strong></td>
<td align="center" style="vertical-align: middle">85.88</td>
<td align="center" style="vertical-align: middle">75.85</td>
</tr>
<tr>
<td align="center" style="vertical-align: middle">GPQA-Diamond</td>
<td align="center" style="vertical-align: middle"><strong>74.43</strong></td>
<td align="center" style="vertical-align: middle">68.69</td>
<td align="center" style="vertical-align: middle">39.90</td>
</tr>
<tr>
<td align="center" style="vertical-align: middle">SuperGPQA</td>
<td align="center" style="vertical-align: middle"><strong>55.00</strong></td>
<td align="center" style="vertical-align: middle">52.00</td>
<td align="center" style="vertical-align: middle">32.00</td>
</tr>
<tr>
<td align="center" style="vertical-align: middle">LiveBench</td>
<td align="center" style="vertical-align: middle"><strong>72.90</strong></td>
<td align="center" style="vertical-align: middle">59.70</td>
<td align="center" style="vertical-align: middle">43.10</td>
</tr>
<tr>
<td align="center" style="vertical-align: middle">IFEval</td>
<td align="center" style="vertical-align: middle"><strong>86.69</strong></td>
<td align="center" style="vertical-align: middle">83.18</td>
<td align="center" style="vertical-align: middle">82.44</td>
</tr>
<tr>
<td align="center" style="vertical-align: middle">AlignBench</td>
<td align="center" style="vertical-align: middle"><strong>8.24</strong></td>
<td align="center" style="vertical-align: middle">8.07</td>
<td align="center" style="vertical-align: middle">6.85</td>
</tr>
<tr>
<td align="center" style="vertical-align: middle">HellaSwag</td>
<td align="center" style="vertical-align: middle"><strong>91.79</strong></td>
<td align="center" style="vertical-align: middle">89.90</td>
<td align="center" style="vertical-align: middle">60.84</td>
</tr>

<tr>
<td align="center" colspan=8><strong>Coding</strong></td>
</tr>
<tr>
<td align="center" style="vertical-align: middle">HumanEval</td>
<td align="center" style="vertical-align: middle"><strong>96.34</strong></td>
<td align="center" style="vertical-align: middle">95.12</td>
<td align="center" style="vertical-align: middle">74.39</td>
</tr>
<tr>
<td align="center" style="vertical-align: middle">LiveCodeBench</td>
<td align="center" style="vertical-align: middle"><strong>65.60</strong></td>
<td align="center" style="vertical-align: middle">39.71</td>
<td align="center" style="vertical-align: middle">27.43</td>
</tr>
<tr>
<td align="center" style="vertical-align: middle">SciCode</td>
<td align="center" style="vertical-align:
middle"><strong>3.08/22.92</strong></td>
<td align="center" style="vertical-align:
middle"><strong>3.08/22.92</strong></td>
<td align="center" style="vertical-align: middle">3.08/15.11</td>
</tr>
<tr>
<td align="center" colspan=8><strong>Mathematics</strong></td>
</tr>
<tr>
<td align="center" style="vertical-align: middle">GSM8K</td>
<td align="center" style="vertical-align: middle"><strong>95.83</strong></td>
<td align="center" style="vertical-align: middle">79.83</td>
<td align="center" style="vertical-align: middle">81.88</td>
</tr>
<tr>
<td align="center" style="vertical-align: middle">AIME2025</td>
<td align="center" style="vertical-align: middle"><strong>65.83</strong></td>
<td align="center" style="vertical-align: middle">62.08</td>
<td align="center" style="vertical-align: middle">24.17</td>
</tr>
<tr>
<td align="center" style="vertical-align: middle">MATH 500</td>
<td align="center" style="vertical-align: middle"><strong>97.10</strong></td>
<td align="center" style="vertical-align: middle">89.80</td>
<td align="center" style="vertical-align: middle">90.90</td>
</tr>

<tr>
<td align="center" colspan=8><strong>Agentic</strong></td>
</tr>
<tr>
<td align="center" style="vertical-align: middle">SWE-bench Verified</td>
<td align="center" style="vertical-align: middle"><strong>60.60</strong></td>
<td align="center" style="vertical-align: middle">24.44</td>
<td align="center" style="vertical-align: middle">51.60</td>
</tr>
<tr>
<td align="center" style="vertical-align: middle">Tau2-Retail</td>
<td align="center" style="vertical-align: middle"><strong>67.55</strong></td>
<td align="center" style="vertical-align: middle">53.51</td>
<td align="center" style="vertical-align: middle">62.28</td>
</tr>
<tr>
<td align="center" style="vertical-align: middle">Tau2-Airline</td>
<td align="center" style="vertical-align: middle"><strong>54.00</strong></td>
<td align="center" style="vertical-align: middle">32.00</td>
<td align="center" style="vertical-align: middle">52.00</td>
</tr>
<tr>
<td align="center" style="vertical-align: middle">Tau2-Telecom</td>
<td align="center" style="vertical-align: middle">79.83</td>
<td align="center" style="vertical-align: middle">4.39</td>
<td align="center" style="vertical-align: middle"><strong>88.60</strong></td>
</tr>

<tr>
<td align="center" colspan=8><strong>Long Context</strong></td>
</tr>
<tr>
<td align="center" style="vertical-align: middle">RULER</td>
<td align="center" style="vertical-align: middle"><strong>95.60</strong></td>
<td align="center" style="vertical-align: middle">89.66</td>
<td align="center" style="vertical-align: middle">56.12</td>
</tr>
</tbody>
</table>


## 4. Deployment

> [!Note]
> You can access JoyAI-LLM Flash API on https://docs.jdcloud.com/cn/jdaip/chat
> and we provide OpenAI/Anthropic-compatible API for you. 
> Currently, JoyAI-LLM Flash is recommended to run on the following inference
> engines:

* vLLM
* SGLang

The minimum version requirement for `transformers` is `4.57.1`.

Deployment examples can be found in the [Model Deployment
Guide](docs/deploy_guidance.md).



## 5. Model Usage

The usage demos below demonstrate how to call our official API.

For third-party APIs deployed with vLLM or SGLang, please note that:

> [!Note] Recommended sampling parameters: `temperature=0.6`, `top_p=1.0`

### Chat Completion

This is a simple chat completion script which shows how to call JoyAI-Flash
API.

```python
from openai import OpenAI

client = OpenAI(base_url="http://IP:PORT/v1", api_key="EMPTY")


def simple_chat(client: OpenAI):
    messages = [
        {
            "role": "user",
            "content": [
                {
                    "type": "text",
                    "text": "which one is bigger, 9.11 or 9.9? think
carefully.",
                }
            ],
        },
    ]
    model_name = client.models.list().data[0].id
    response = client.chat.completions.create(
        model=model_name, messages=messages, stream=False, max_tokens=4096
    )
    print(f"response: {response.choices[0].message.content}")


if __name__ == "__main__":
    simple_chat(client)
```


### Tool call Completion

This is a simple toll call completion script which shows how to call
JoyAI-Flash API.

```python
import json

from openai import OpenAI

client = OpenAI(base_url="http://IP:PORT/v1", api_key="EMPTY")


def my_calculator(expression: str) -> str:
    return str(eval(expression))


def rewrite(expression: str) -> str:
    return str(expression)


def simple_tool_call(client: OpenAI):
    messages = [
        {
            "role": "user",
            "content": [
                {
                    "type": "text",
                    "text": "use my functions to compute the results for the
equations: 6+1",
                },
            ],
        },
    ]
    tools = [
        {
            "type": "function",
            "function": {
                "name": "my_calculator",
                "description": "A calculator that can evaluate a mathematical
equation and compute its results.",
                "parameters": {
                    "type": "object",
                    "properties": {
                        "expression": {
                            "type": "string",
                            "description": "The mathematical expression to
evaluate.",
                        },
                    },
                    "required": ["expression"],
                },
            },
        },
        {
            "type": "function",
            "function": {
                "name": "rewrite",
                "description": "Rewrite a given text for improved clarity",
                "parameters": {
                    "type": "object",
                    "properties": {
                        "text": {
                            "type": "string",
                            "description": "The input text to rewrite",
                        }
                    },
                },
            },
        },
    ]
    model_name = client.models.list().data[0].id
    response = client.chat.completions.create(
        model=model_name,
        messages=messages,
        temperature=1.0,
        max_tokens=1024,
        tools=tools,
        tool_choice="auto",
    )
    tool_calls = response.choices[0].message.tool_calls

    results = []
    for tool_call in tool_calls:
        function_name = tool_call.function.name
        function_args = tool_call.function.arguments
        if function_name == "my_calculator":
            result = my_calculator(**json.loads(function_args))
            results.append(result)
    messages.append({"role": "assistant", "tool_calls": tool_calls})
    for tool_call, result in zip(tool_calls, results):
        messages.append(
            {
                "role": "tool",
                "tool_call_id": tool_call.id,
                "name": tool_call.function.name,
                "content": result,
            }
        )
    response = client.chat.completions.create(
        model=model_name,
        messages=messages,
        temperature=1.0,
        max_tokens=1024,
    )
    print(response.choices[0].message.content)


if __name__ == "__main__":
    simple_tool_call(client)

```

---

## 6. License

Both the code repository and the model weights are released under the [Modified
MIT License](LICENSE).