MobileLLM-R1

April 30, 2026 · View on GitHub

MobileLLM-R1

MobileLLM-R1 is a new series of efficient reasoning models within the MobileLLM family. Alongside the models, we provide comprehensive training recipes and data sources to ensure reproducibility and facilitate further research.

This repository includes demonstration code to reproduce the pretraining, mid-training, and SFT stages of MobileLLM-R1, as well as the corresponding intermediate checkpoints and data mix weights.

News

Jan 22, 2026: 🌟 MobileLLM-R1 has been accepted to ICLR 2026.
Sept 29, 2025: 🌟 The technical report "MobileLLM-R1: Exploring the Limits of Sub-Billion Language Model Reasoners with Open Training Recipes" is also available! Please check it out.
Sept 12, 2025: 🚀 MobileLLM-R1 models are released on HuggingFace.

Highlights

Remarkably, the MobileLLM-R1 950M, pre-trained on only ~2T high-quality tokens and with fewer than 5T total training tokens, achieves comparable or superior performance to Qwen3 0.6B, which was trained on 36T tokens, across MATH, GSM8K, MMLU, and LiveCodeBench benchmarks.

Compared to existing fully open-source models, MobileLLM-R1 950M model achieves ~5× higher accuracy on MATH compared to the Olmo 1.24B model and ~2× higher accuracy relative to the SmolLM2 1.7B model, despite being substantially smaller in parameter scale. In addition, MobileLLM-R1 950M outperforms both Olmo 1.24B and SmolLM2 1.7B by a wide margin on coding benchmarks, establishing a new state-of-the-art among fully open-source models.

Pretrained Model

image/jpeg

Token efficiency comparison across pretrained models

Post-trained Model

image/jpeg

Model Architecture

	# Layers	# Attnetion Heads	# KV Heads	Dim	Hidden Dim	Params
MobileLLM-R1-140M	15	9	3	576	2048	140M
MobileLLM-R1-360M	15	16	4	1024	4096	359M
MobileLLM-R1-950M	22	24	6	1536	6144	949M

	Input modalities	Output modalities	Context Length	Vocaburary Size	Shared Embeddings
MobileLLM-R1-140M-base	Text	Text	4k	128k	Yes
MobileLLM-R1-360M-base	Text	Text	4k	128k	Yes
MobileLLM-R1-950M-base	Text	Text	4k	128k	Yes
MobileLLM-R1-140M	Text	Text	32k	128k	Yes
MobileLLM-R1-360M	Text	Text	32k	128k	Yes
MobileLLM-R1-950M	Text	Text	32k	128k	Yes

Training

Training Process

image/jpeg

Training stages and hyperparameter details

In the pretraining phase, MobileLLM-R1 models are randomly initialized and optimized using the Adam optimizer with hyperparameters (β_1, β_2, ε) = (0.9, 0.95, 1e-8), coupled with a weight decay coefficient of 0.1. The learning rate follows a 2k-step warmup schedule and then decays linearly from its peak to 10% of the maximum.

In the mid-training phase, we use Adam optimizer with learning rate linearly decays from its maximum value to zero. We employ knowledge distillation with Llama-3.1-8B-Instruct model as the teacher, where the student is trained via minimizing the KL divergence between its output logits and the teacher logits.

In the post-training phase, we use the Adam optimizer with zero weight decay. The learning rate warmup ratio is set to 0.03 for general-purpose SFT and 0.1 for reasoning-specific SFT, and it linearly decays from its maximum value to zero. Full training hyperparameters are provided in the table below.

Stage	Phase	Tokens / Samples	BS	Sequence Length	Steps	LR	#GPUs	Training Time
Pre-training	Phase1	2T tokens	16	2k	500k	4.00E-03	16 x 8	4-5 days
	Phase2	2T tokens	16	2k	500k	4.00E-03	16 x 8	4-5 days
Mid-training	Phase1	100B tokens	4	4k	50K	3.60E-04	16 x 8	1-2 days
	Phase2	100B tokens	4	4k	50K	3.60E-04	16 x 8	1-2 days
Post-training	General SFT	866K samples	4	4k	2 epochs	5.00E-06	16 x 8	~2h
	Reasoning SFT	6.2M samples	8	32k	4 epochs	8.00E-05	16 x 8	~2.5days

Quick Start

To load the pretrained model for further fine-tuning or evaluation:

from transformers import AutoModelForCausalLM, AutoTokenizer
tokenizer = AutoTokenizer.from_pretrained("facebook/MobileLLM-R1-950M")
model = AutoModelForCausalLM.from_pretrained("facebook/MobileLLM-R1-950M")

Pretraining

Training code

The pretraining code that includes support for data mixing is available under ./pretraining. Example code:

cd ./pretraining
bash run_pretrain.sh

Note that our implementation does not include efficiency optimizations; for training speedups, you may refer to open-source efficiency implementations such as torchtitan or torchtune.

Intermediate checkpoints

Model Size	Pretraining Stage1	Pretraining Stage2	Intermediate Checkpoints*
950M	MobileLLM-R1-950M-base	MobileLLM-R1-950M-base	Coming Soon*
360M	MobileLLM-R1-360M-base	MobileLLM-R1-360M-base	Coming Soon*
140M	MobileLLM-R1-140M-base	MobileLLM-R1-140M-base	Coming Soon*

*Links to intermediate checkpoints will be made available soon.

Mid-training

Training code

Mid-training only differs from pre-training in the dataset selection and training steps. The training code structure are identical to pretraining. Please refer to the ./pretraining section for details on the training code.

Intermediate checkpoints

Model Size	Mid-training Stage1	Mid-training Stage2	Intermediate Checkpoints*
950M	MobileLLM-R1-950M-base	MobileLLM-R1-950M-base	Coming Soon*
360M	MobileLLM-R1-360M-base	MobileLLM-R1-360M-base	Coming Soon*
140M	MobileLLM-R1-140M-base	MobileLLM-R1-140M-base	Coming Soon*

*Links to intermediate checkpoints will be made available soon.

Post-training

Training code

Training code for the post-training stage is available under ./sft. The code is based on TRL. Example code:

cd ./sft
bash run_general_sft.sh
bash run_reasoning_sft.sh

Intermediate checkpoints

Model Size	General SFT	Reasoning SFT
950M	MobileLLM-R1-950M	MobileLLM-R1-950M
360M	MobileLLM-R1-360M	MobileLLM-R1-360M
140M	MobileLLM-R1-140M	MobileLLM-R1-140M

Inference

Inference examples

Transformers

from transformers import pipeline
import torch

model_id = "facebook/MobileLLM-R1-950M"

pipe = pipeline(
    "text-generation",
    model=model_id,
    torch_dtype="auto",
    device_map="auto",
)

# Math problem / default scenario
messages = [
    {
        "role": "system",
        "content": "Please reason step by step, and put your final answer within \\boxed{}."
    },
    {"role": "user", "content": "Compute: \$1-2+3-4+5- \\dots +99-100$."},
]

# C++ coding scenario
messages = [
    {
        "role": "system",
        "content": (
            "\nYou are a helpful and harmless assistant. You should think step-by-step before responding to the instruction below.\n\n"
            "Please use c++ programming language only.\n"
            "You must use ```cpp for just the final solution code block with the following format:\n"
            "```cpp\n# Your code here\n```\n"
        )
    },
    {"role": "user", "content": "Write a C++ program that prints 'Hello, World!'."},
]

# Python coding scenario
messages = [
    {
        "role": "system",
        "content": (
            "\nYou are a helpful and harmless assistant. You should think step-by-step before responding to the instruction below.\n\n"
            "Please use python programming language only.\n"
            "You must use ```python for just the final solution code block with the following format:\n"
            "```python\n# Your code here\n```\n"
        )
    },
    {"role": "user", "content": "Write a Python function that returns the square of a number."},
]

outputs = pipe(
    messages,
    max_new_tokens=8192,
)
print(outputs[0]["generated_text"][-1])

You can also run inference with vLLM. You only need to register the model architecture Llama4ForCausalLM with the vLLM ModelRegistry.

from vllm.model_executor.models.llama4 import Llama4ForCausalLM
from vllm.model_executor.models.registry import ModelRegistry
ModelRegistry.register_model("Llama4ForCausalLM", Llama4ForCausalLM)

Data mix

Pretraining

Dataset	Rows	Tokens (B)	Phase1 Mix Ratio	Phase2 Mix Ratio
StarCoder	206,640,114	263.8	10.66%	0.52%
OpenWebMath	6,117,786	12.6	6.93%	23.33%
FineWeb-Edu	1,279,107,432	1300	63.75%	54.83%
Wiki	7,222,303	3.7	5.03%	0.14%
Arxiv	1,533,917	28	6.36%	1.32%
StackExchange	29,249,120	19.6	5.03%	0.86%
Algebraic stack	3,404,331	12.6	2.25%	1.26%
Nemotron science	708,920	2	--	0.03%
Nemotron code	10,108,883	16	--	0.72%
Nemotron math	22,066,397	15	--	3.01%
Cosmopedia	31,064,744	25	--	2.70%
Facebook natural reasoning	1,145,824	1.8	--	3.18%
FineMath	48,283,984	34	--	8.01%
peS2o	38,800,000	50	--	0.08%
Total			100%	100%

Mid-training

Dataset	Subset	Rows (M)	Phase1 Mix Ratio	Phase2 Mix Ratio
Dolmino	DCLM Baseline	606	37.03%	6.51%
	FLAN	57.3	4.10%	0.72%
	peS2o	38.8	11.41%	2.01%
	Wiki	6.17	2.66%	0.47%
	StackExchange	2.48	2.12%	2.00%
	Math	21	11.63%	29.10%
Nemotron	Nemotron-Pretraining-Code-v1	882	20.69%	29.10%
	Nemotron-CC-Math-v1	144	3.45%	19.40%
StarCoder	StarCoder	206	6.90%	9.70%
Benchmark training set	TriviaQA (train) OBQA (train) NaturalQuestions (train) PIQA (train) GSM8K (train) BoolQ (train) ARC-Easy (train) ARC-Challenge (train)	~0.01	0	0.97%
Total			100.00%	100.00%

Post-training

Phase	Dataset	Rows
General SFT	Tulu-3-sft-olmo-2-mixture-0225	866K samples
Reasoning SFT	OpenMathReasoning	3.2M samples
	OpenScienceReasoning-2	803K samples
	OpenCodeReasoning-2	2.16M samples

Evaluation

Evaluation code

We provide the evaluation code necessary to reproduce the MobileLLM-R1 evaluation results in the evaluation folder. The evaluation results are summarized in the following table.

MobileLLM-R1 base model

Model	Size	MATH500	GSM8K	MBPP	HumanEval	CommonSense Avg.	MMLU
		4-shot em	8-shot em	3-shot pass@1	0-shot pass@1	0-shot accuracy	5-shot accuracy

<150M
SmolLM2-135M-base	135M	0.4	1.8	3.8	0.0	50.7	--
MobileLLM-R1-140M-base	140M	4.6	16.3	5.4	15.9	44.3	--

150M - 400M
Gemma-3-270M-pt	268M	0.6	1.1	2.0	3.1	48.4	26.5
SmolLM2-360M-base	362M	1.8	5.0	19.4	0.0	56.6	24.7
MobileLLM-R1-360M-base	359M	13.4	39.4	20.8	32.9	51.0	26.8

400M - 1B
Qwen2.5-0.5B-base	494M	14.8	41.8	29.6	28.1	52.3	47.5
Qwen3-0.6B-base	596M	29.8	60.9	39.0	30.5	55.3	52.4
MobileLLM-R1-950M-base	949M	26.8	61.6	39.2	46.3	58.6	47.4

> 1B
Gemma-3-1B-pt	1.0B	0.6	2.4	9.4	6.1	57.3	26.1
LLaMA3.2-1B-base	1.24B	1.6	6.8	26.6	17.1	58.4	32.0
OLMo-2-0425-1B-base	1.48B	5.2	39.8	7.8	6.7	61.0	42.4
Qwen2.5-1.5B-base	1.54B	31.0	68.4	44.6	36.6	58.7	61.2
SmolLM2-1.7B-base	1.71B	11.6	31.8	35.4	0.6	62.9	50.0
Qwen3-1.7B-base	2.03B	38.5	76.2	56.4	47.6	60.9	62.1

Here, CommonSense Avg. denotes an average of 8 tasks in CommonSense Reasoning benchmarks including ARC-easy, ARC-challenge, BoolQ, PIQA, SIQA, HellaSwag, OBQA, and WinoGrand. Models with fewer than 150M parameters do not yield reliable MMLU scores and are therefore denoted as '—'.

MobileLLM-R1 post-trained model

Model	Size	MATH500	GSM8K	AIME'24	AIME'25	LiveCodeBench-v6
		0-shot pass@1	0-shot pass@1	0-shot pass@1, n=64	0-shot pass@1, n=64	0-shot pass@1, n=16

<150M
SmolLM2-135M-Instruct	135M	3.0	2.4	--	--	0.0
MobileLLM-R1-140M	140M	6.2	4.1	--	--	1.7

150M - 400M
Gemma-3-270m-it	268M	6.8	8.4	--	--	0.0
SmolLM2-360M-Instruct	362M	3.4	8.1	--	--	0.7
MobileLLM-R1-360M	359M	28.4	24.5	--	--	5.1

400M - 1B
Qwen2.5-0.5B-Instruct	494M	31.2	48.1	0.1	0.3	3.6
Qwen3-0.6B	596M	73.0	79.2	11.3	17.0	14.9
MobileLLM-R1-950M	949M	74.0	67.5	15.5	16.3	19.9

> 1B
Gemma-3-1B-it	1.0B	45.4	62.9	0.9	0.0	2.0
LLaMA3.2-1B-Instruct	1.24B	24.8	38.8	1.1	0.2	4.1
OLMo-2-0425-1B-Instruct	1.48B	19.2	69.7	0.6	0.1	0.0
OpenReasoning-Nemotron-1.5B	1.54B	83.4	76.7	49.7	40.4	28.3
DeepSeek-R1-Distill-Qwen-1.5B	1.54B	83.2	77.3	29.1	23.4	19.9
Qwen2.5-1.5B-Instruct	1.54B	54.0	70.0	2.5	0.9	7.9
SmolLM2-1.7B-Instruct	1.71B	19.2	41.8	0.3	0.1	4.4
Qwen3-1.7B	2.03B	89.4	90.3	47.0	37.0	29.8

For AIME, we evaluate models across 64 runs and report the average accuracy. For LiveCodeBench, results are reported as the average accuracy across 16 runs. Models with fewer than 400M parameters do not produce reliable AIME scores and are therefore denoted as '—'.

Citation

If you find our model useful for your research, please consider citing:

@article{zhao2025mobilellm-r1,
  title={MobileLLM-R1: Exploring the Limits of Sub-Billion Language Model Reasoners with Open Training Recipes},
  author={Zhao, Changsheng and Chang, Ernie and Liu, Zechun and Chang, Chia-Jung and Wen, Wei and Lai, Chen and Cao, Sheng, and Tian, Yuandong and Krishnamoorthi, Raghuraman and Shi, Yangyang and  Chandra, Vikas},
  journal={arXiv preprint arXiv:2509.24945},
  year={2025}
}

Contact

Changsheng Zhao, Meta Inc (cszhao at meta dot com)

Ernie Chang, Meta Inc (erniecyc at meta dot com)

Zechun Liu, Meta Inc (zechunliu at meta dot com)

License

MobileLLM-R1 is FAIR NC licensed as of now