library_name: transformers
license: apache-2.0
license_link: https://huggingface.co/internlm/Intern-S2-Preview/blob/main/LICENSE
pipeline_tag: image-text-to-text
Intern-S2-Preview
👋 join us on Discord and WeChat
Introduction
We introduce Intern-S2-Preview, an efficient 35B scientific multimodal foundation model. Beyond conventional parameter and data scaling, Intern-S2-Preview explores task scaling: increasing the difficulty, diversity, and coverage of scientific tasks to further unlock model capabilities.
By extending professional scientific tasks into a full-chain training pipeline from pre-training to reinforcement learning, Intern-S2-Preview achieves performance comparable to the trillion-scale Intern-S1-Pro on multiple core professional scientific tasks, while using only 35B parameters. At the same time, it maintains strong general reasoning, multimodal understanding, coding, and agent capabilities.
Features
Scientific task scaling with full-chain training. Intern-S2-Preview scales hundreds of professional scientific tasks from pre-training to RL, enabling strong performance across multiple specialized domains at only 35B parameters. It further strengthens spatial modeling for small-molecule structures and introduces real-valued prediction modules, making it the first open-source model with both material crystal structure generation capability and strong general capabilities.
Enhanced agent capabilities for scientific workflows. Intern-S2-Preview significantly improves agentic abilities over the previous generation, achieving strong results on multiple scientific agent benchmarks.
Efficient RL reasoning with MTP and CoT compression. During RL, Intern-S2-Preview adopts shared-weight MTP with KL loss to reduce the mismatch between training and inference behavior, substantially improving MTP accept rate and token generation speed. It also introduces CoT compression techniques to shorten responses while preserving strong reasoning capability, achieving improvements in both performance and efficiency.
Performance
We evaluate the Intern-S2-Preview on various benchmarks, including general datasets and scientific datasets. We report the performance comparison with the recent VLMs and LLMs below.
Note: Underline means the best performance among open-sourced models, Bold indicates the best performance among all models.
We use the OpenCompass and VLMEvalKit to evaluate all models.
Quick Start
Sampling Parameters
We recommend using the following hyperparameters to ensure better results
top_p = 0.95
top_k = 50
min_p = 0.0
temperature = 0.8
Serving
Intern-S2-Preview can be deployed using any of the following LLM inference frameworks:
- LMDeploy
- vLLM
- SGLang
Detailed deployment examples for these frameworks are available in the Model Deployment Guide.
Advanced Usage
Tool Calling
Tool Calling lets the model extend its capabilities by invoking external tools and APIs. The example below shows how to use it to fetch the latest weather forecast via an OpenAI-compatible API (based on lmdeploy api server).
from openai import OpenAI
import json
def get_current_temperature(location: str, unit: str = "celsius"):
"""Get current temperature at a location.
Args:
location: The location to get the temperature for, in the format "City, State, Country".
unit: The unit to return the temperature in. Defaults to "celsius". (choices: ["celsius", "fahrenheit"])
Returns:
the temperature, the location, and the unit in a dict
"""
return {
"temperature": 26.1,
"location": location,
"unit": unit,
}
def get_temperature_date(location: str, date: str, unit: str = "celsius"):
"""Get temperature at a location and date.
Args:
location: The location to get the temperature for, in the format "City, State, Country".
date: The date to get the temperature for, in the format "Year-Month-Day".
unit: The unit to return the temperature in. Defaults to "celsius". (choices: ["celsius", "fahrenheit"])
Returns:
the temperature, the location, the date and the unit in a dict
"""
return {
"temperature": 25.9,
"location": location,
"date": date,
"unit": unit,
}
def get_function_by_name(name):
if name == "get_current_temperature":
return get_current_temperature
if name == "get_temperature_date":
return get_temperature_date
tools = [{
'type': 'function',
'function': {
'name': 'get_current_temperature',
'description': 'Get current temperature at a location.',
'parameters': {
'type': 'object',
'properties': {
'location': {
'type': 'string',
'description': 'The location to get the temperature for, in the format \'City, State, Country\'.'
},
'unit': {
'type': 'string',
'enum': [
'celsius',
'fahrenheit'
],
'description': 'The unit to return the temperature in. Defaults to \'celsius\'.'
}
},
'required': [
'location'
]
}
}
}, {
'type': 'function',
'function': {
'name': 'get_temperature_date',
'description': 'Get temperature at a location and date.',
'parameters': {
'type': 'object',
'properties': {
'location': {
'type': 'string',
'description': 'The location to get the temperature for, in the format \'City, State, Country\'.'
},
'date': {
'type': 'string',
'description': 'The date to get the temperature for, in the format \'Year-Month-Day\'.'
},
'unit': {
'type': 'string',
'enum': [
'celsius',
'fahrenheit'
],
'description': 'The unit to return the temperature in. Defaults to \'celsius\'.'
}
},
'required': [
'location',
'date'
]
}
}
}]
messages = [
{'role': 'user', 'content': 'Today is 2024-11-14, What\'s the temperature in San Francisco now? How about tomorrow?'}
]
openai_api_key = "EMPTY"
openai_api_base = "http://0.0.0.0:23333/v1"
client = OpenAI(
api_key=openai_api_key,
base_url=openai_api_base,
)
model_name = client.models.list().data[0].id
response = client.chat.completions.create(
model=model_name,
messages=messages,
max_tokens=32768,
temperature=0.8,
top_p=0.95,
extra_body=dict(spaces_between_special_tokens=False),
tools=tools)
print(response.choices[0].message)
messages.append(response.choices[0].message)
for tool_call in response.choices[0].message.tool_calls:
tool_call_args = json.loads(tool_call.function.arguments)
tool_call_result = get_function_by_name(tool_call.function.name)(**tool_call_args)
tool_call_result = json.dumps(tool_call_result, ensure_ascii=False)
messages.append({
'role': 'tool',
'name': tool_call.function.name,
'content': tool_call_result,
'tool_call_id': tool_call.id
})
response = client.chat.completions.create(
model=model_name,
messages=messages,
temperature=0.8,
top_p=0.95,
extra_body=dict(spaces_between_special_tokens=False),
tools=tools)
print(response.choices[0].message)
Switching Between Thinking and Non-Thinking Modes
Intern-S2-Preview enables thinking mode by default, enhancing the model's reasoning capabilities to generate higher-quality responses. This feature can be disabled by setting enable_thinking=False in tokenizer.apply_chat_template
text = tokenizer.apply_chat_template(
messages,
tokenize=False,
add_generation_prompt=True,
enable_thinking=False # think mode indicator
)
When serving Intern-S2-Preview models, you can dynamically control the thinking mode by adjusting the enable_thinking parameter in your requests.
from openai import OpenAI
import json
messages = [
{
'role': 'user',
'content': 'who are you'
}, {
'role': 'assistant',
'content': 'I am an AI'
}, {
'role': 'user',
'content': 'AGI is?'
}]
openai_api_key = "EMPTY"
openai_api_base = "http://0.0.0.0:23333/v1"
client = OpenAI(
api_key=openai_api_key,
base_url=openai_api_base,
)
model_name = client.models.list().data[0].id
response = client.chat.completions.create(
model=model_name,
messages=messages,
temperature=0.8,
top_p=0.95,
max_tokens=2048,
extra_body={
"chat_template_kwargs": {"enable_thinking": False}
}
)
print(json.dumps(response.model_dump(), indent=2, ensure_ascii=False))
Note: We do not recommend disabling thinking mode for agentic tasks.