LLaDA2.0-flash

LLaDA2.0-flash is a diffusion language model featuring a 100BA6B Mixture-of-Experts (MoE) architecture. As an enhanced, instruction-tuned iteration of the LLaDA2.0 series, it is optimized for practical applications.

---

Benchmark	Qwen3-30B-A3B-Instruct-2507	Ling-flash-2.0	LLaDA2.0-flash-preview	LLaDA2.0-flash
Average	79.47	78.03	71.92	79.32
Knowledge
MMLU	87.13	87.98	83.15	87.69
MMLU-Pro	74.23	76.84	49.22	73.36
GPQA	57.34	67.12	46.59	61.98
arc-c	95.81	95.08	93.90	95.93
CMMLU	86.36	86.59	67.53	85.13
C-EVAL	88.17	88.03	66.54	86.75
GAOKAO-Bench	94.53	93.24	86.12	93.90
Reasoning
SQuAD 2.0	89.51	81.32	85.61	90.00
DROP	87.57	88.32	79.49	87.90
KOR-Bench	68.00	68.96	37.26	64.24
HellaSwag	86.31	81.59	86.00	84.97
Coding
CRUXEval-O	86.75	82.75	61.88	85.12
MBPP	86.65	85.01	77.75	88.29
MultiPL-E	70.67	65.76	62.43	74.87
HumanEval	93.29	85.98	80.49	94.51
Bigcodebench-Full	41.49	40.70	30.44	41.58
LiveCodeBench	41.63	44.11	28.58	42.29
Spider	81.79	80.58	81.37	82.49
Math
GSM8K	96.36	95.45	89.01	96.06
MATH	96.70	96.1	73.50	95.44
OlympiadBench	77.59	76.19	47.78	74.07
AIME 2025	61.88	55.89	23.33	60.00
Agent & Alignment
BFCL_Live	73.19	67.57	74.11	75.43
IFEval-strict -prompt	84.29	81.52	62.50	81.70

🚀 Performance Highlights

• Leading MoE Architecture: The open-source Mixture-of-Experts (MoE) diffusion large language model continually trained on the Ling2.0 series with approximately 20 trillion tokens.
• Efficient Inference: With 100 billion total parameters, only 6.1 billion are activated during inference. LLaDA2.0-flash significantly reduces computational costs while outperforming open-source dense models of similar scale.
• Impressive Performance on Code & Complex Reasoning: Excels in tasks such as code generation and advanced mathematical reasoning, demonstrating strong reasoning capabilities.
• Tool Use: Supports tool calling and achieves excellent performance in complex agent-based tasks.
• Open & Extensible: Fully open-source with commitment to transparency. We plan to release a leading inference framework in the future and continue investing in cutting-edge areas like diffusion LLMs (dLLM) to drive disruptive innovation.

🗺️ What's Next

• Supercharged Reasoning with LLaDA 2.0: LLaDA 2.0 series will be fine-tuned with Reinforcement Learning, unlocking a new level of sophisticated reasoning and problem-solving abilities.
• Tools for Innovators: The model was finetuned on the dFactory framework using Fully Sharded Data Parallel (FSDP2). We have begun open-sourcing dFactory and will continuously release our advanced post-training technologies. Whether you want to master the current model or build your own customized versions, you'll have the tools you need. Stay tuned for more updates!

---

📦 Model Variants

Model ID	Description	Hugging Face Link
inclusionAI/LLaDA2.0-mini	Instruction-tuned model, ready for downstream applications.	🤗 Model Card
inclusionAI/LLaDA2.0-flash	Instruction-tuned model, ready for downstream applications.	🤗 Model Card

---

🔍 Model Overview

LLaDA2.0-flash has the following specifications:

• Type: Mixture-of-Experts (MoE) Diffusion Language Model
• Total Parameters (Non-Embedding): 100B
• Number of Layers: 32
• Attention Heads: 32
• Context Length: 32,768 tokens
• Position Embedding: Rotary (RoPE)
• Vocabulary Size: 157,184

---

🤗 Hugging Face Transformers

Make sure you have transformers and its dependencies installed:

import torch
import torch.nn.functional as F
from transformers import AutoModelForCausalLM
from transformers import AutoTokenizer

model_path = "/path/to/LLaDA2.0-mini-preview"
device = "auto"
model = AutoModelForCausalLM.from_pretrained(
    model_path, trust_remote_code=True, device_map=device
)
model = model.to(torch.bfloat16)
model.eval()
tokenizer = AutoTokenizer.from_pretrained(model_path, trust_remote_code=True)

prompt = "Why does Camus think that Sisyphus is happy?"
input_ids = tokenizer.apply_chat_template(
    [{"role": "user", "content": prompt}],
    add_generation_prompt=True,
    tokenize=True,
    return_tensors="pt",
)
generated_tokens = model.generate(
    inputs=input_ids,
    eos_early_stop=True,
    gen_length=512,
    block_length=32,
    steps=32,
    temperature=0.0,
)
generated_answer = tokenizer.decode(
    generated_tokens[0],
    skip_special_tokens=True,
)
print(generated_answer)

Best Practices

To achieve optimal performance, we recommend the following settings:

1. Sampling Parameters: We suggest using Temperature=0.0, block_length=32, and steps=32. Using a higher temperature value may occasionally result in language mixing and a slight decrease in model performance.

2. Adequate Output Length: We recommend using an output length of 32768 tokens for most queries.

---

🌐 License

This project is licensed under the terms of the Apache License 2.0.

---

🤝 Contact & Collaboration

For questions, collaborations, or feedback, please reach out via Hugging Face or open an issue in the repository.

👉 Join us in advancing open, efficient, and intelligent language models!