Instructions to use fangwu97/DeepSearch-1.5B with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use fangwu97/DeepSearch-1.5B with Transformers:

# Use a pipeline as a high-level helper
from transformers import pipeline

pipe = pipeline("text-generation", model="fangwu97/DeepSearch-1.5B")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("fangwu97/DeepSearch-1.5B")
model = AutoModelForCausalLM.from_pretrained("fangwu97/DeepSearch-1.5B")
messages = [
    {"role": "user", "content": "Who are you?"},
]
inputs = tokenizer.apply_chat_template(
	messages,
	add_generation_prompt=True,
	tokenize=True,
	return_dict=True,
	return_tensors="pt",
).to(model.device)

outputs = model.generate(**inputs, max_new_tokens=40)
print(tokenizer.decode(outputs[0][inputs["input_ids"].shape[-1]:]))

Notebooks
Google Colab
Kaggle
Local Apps

vLLM

How to use fangwu97/DeepSearch-1.5B with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "fangwu97/DeepSearch-1.5B"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "fangwu97/DeepSearch-1.5B",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/fangwu97/DeepSearch-1.5B

SGLang

How to use fangwu97/DeepSearch-1.5B with SGLang:

Install from pip and serve model

# Install SGLang from pip:
pip install sglang
# Start the SGLang server:
python3 -m sglang.launch_server \
    --model-path "fangwu97/DeepSearch-1.5B" \
    --host 0.0.0.0 \
    --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "fangwu97/DeepSearch-1.5B",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker images

docker run --gpus all \
    --shm-size 32g \
    -p 30000:30000 \
    -v ~/.cache/huggingface:/root/.cache/huggingface \
    --env "HF_TOKEN=<secret>" \
    --ipc=host \
    lmsysorg/sglang:latest \
    python3 -m sglang.launch_server \
        --model-path "fangwu97/DeepSearch-1.5B" \
        --host 0.0.0.0 \
        --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "fangwu97/DeepSearch-1.5B",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use fangwu97/DeepSearch-1.5B with Docker Model Runner:
```
docker model run hf.co/fangwu97/DeepSearch-1.5B
```

Add pipeline tag and hyperlink paper in model card

by nielsr HF Staff - opened Oct 3, 2025

base: refs/heads/main

←

from: refs/pr/1

Discussion Files changed

+12

-43

Files changed (1) hide show

README.md +12 -43

README.md CHANGED Viewed

@@ -1,7 +1,11 @@
 ---
 language:
 - en
 library_name: transformers
 tags:
 - reasoning
 - reinforcement-learning
@@ -9,67 +13,30 @@ tags:
 - mcts
 - math
 - iclr-2026
-license: apache-2.0
-datasets:
-- DeepMath-103K
 model-index:
 - name: DeepSearch-1.5B
   results:
   - task:
-      name: Mathematical Reasoning
       type: text-generation
     dataset:
       name: AIME 2024
       type: text
     metrics:
     - type: avg@32
       value: 53.65
-  - task:
-      name: Mathematical Reasoning
-      type: text-generation
-    dataset:
-      name: AIME 2025
-      type: text
-    metrics:
     - type: avg@32
       value: 35.42
-  - task:
-      name: Mathematical Reasoning
-      type: text-generation
-    dataset:
-      name: AMC 2023
-      type: text
-    metrics:
     - type: avg@32
       value: 90.39
-  - task:
-      name: Mathematical Reasoning
-      type: text-generation
-    dataset:
-      name: MATH500
-      type: text
-    metrics:
     - type: avg@32
       value: 92.53
-  - task:
-      name: Mathematical Reasoning
-      type: text-generation
-    dataset:
-      name: Minerva
-      type: text
-    metrics:
     - type: avg@32
-      value: 40.00
-  - task:
-      name: Mathematical Reasoning
-      type: text-generation
-    dataset:
-      name: Olympiad
-      type: text
-    metrics:
     - type: avg@32
       value: 65.72
 ---
 <div align="center">
 <span style="font-family: default; font-size: 1.5em;">🚀 DeepSearch-1.5B</span>
 </div>
@@ -88,7 +55,7 @@ This model achieves **state-of-the-art accuracy among 1.5B reasoning models** wh
 - **Developed by**: Fang Wu\*, Weihao Xuan\*, Heli Qi\*, Ximing Lu, Aaron Tu, Li Erran Li, Yejin Choi
 - **Institutional affiliations**: Stanford University, University of Tokyo, RIKEN AIP, University of Washington, UC Berkeley, Amazon AWS, Columbia University
-- **Paper**: DeepSearch: Overcome the Bottleneck of Reinforcement Learning with Verifiable Rewards via Monte Carlo Tree Search
 - **Base Model**: Nemotron-Research-Reasoning-Qwen-1.5B v2
 - **Parameters**: 1.5B
 - **Framework**: veRL
@@ -114,7 +81,8 @@ from transformers import AutoTokenizer
 def convert_question_to_messages(question: str):
     messages = [
         {"role": "user",
-         "content": question + " Let's think step by step and output the final answer within \\boxed{}."}
     ]
     return messages
@@ -155,7 +123,7 @@ print(response)
 | Olympiad  | 64.69 | **65.72** |
 | **Average** | 61.70 | **62.95** |
-DeepSearch improves average accuracy by **+1.25 points** over the best prior 1.5B model, while using **5.7× fewer GPU hours**.
 ## Training
@@ -191,3 +159,4 @@ DeepSearch improves average accuracy by **+1.25 points** over the best prior 1.5
   primaryClass = {cs.AI},
   doi          = {10.48550/arXiv.2509.25454},
 }

 ---
+datasets:
+- DeepMath-103K
 language:
 - en
 library_name: transformers
+license: apache-2.0
+pipeline_tag: text-generation
 tags:
 - reasoning
 - reinforcement-learning
 - mcts
 - math
 - iclr-2026
 model-index:
 - name: DeepSearch-1.5B
   results:
   - task:
       type: text-generation
+      name: Mathematical Reasoning
     dataset:
       name: AIME 2024
       type: text
     metrics:
     - type: avg@32
       value: 53.65
     - type: avg@32
       value: 35.42
     - type: avg@32
       value: 90.39
     - type: avg@32
       value: 92.53
     - type: avg@32
+      value: 40.0
     - type: avg@32
       value: 65.72
 ---
 <div align="center">
 <span style="font-family: default; font-size: 1.5em;">🚀 DeepSearch-1.5B</span>
 </div>
 - **Developed by**: Fang Wu\*, Weihao Xuan\*, Heli Qi\*, Ximing Lu, Aaron Tu, Li Erran Li, Yejin Choi
 - **Institutional affiliations**: Stanford University, University of Tokyo, RIKEN AIP, University of Washington, UC Berkeley, Amazon AWS, Columbia University
+- **Paper**: [DeepSearch: Overcome the Bottleneck of Reinforcement Learning with Verifiable Rewards via Monte Carlo Tree Search](https://huggingface.co/papers/2509.25454)
 - **Base Model**: Nemotron-Research-Reasoning-Qwen-1.5B v2
 - **Parameters**: 1.5B
 - **Framework**: veRL
 def convert_question_to_messages(question: str):
     messages = [
         {"role": "user",
+         "content": question + " Let's think step by step and output the final answer within \\boxed{}. \
+"}
     ]
     return messages
 | Olympiad  | 64.69 | **65.72** |
 | **Average** | 61.70 | **62.95** |
+DeepSearch improves average accuracy by **+1.25 points** over the best prior 1.5B model, while using **5.7× more GPU hours**.
 ## Training
   primaryClass = {cs.AI},
   doi          = {10.48550/arXiv.2509.25454},
 }
+```