Instructions to use InternScience/Agents-K1 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use InternScience/Agents-K1 with Transformers:

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

pipe = pipeline("text-generation", model="InternScience/Agents-K1")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoTokenizer, AutoModelForMultimodalLM

tokenizer = AutoTokenizer.from_pretrained("InternScience/Agents-K1")
model = AutoModelForMultimodalLM.from_pretrained("InternScience/Agents-K1")
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 Settings

vLLM

How to use InternScience/Agents-K1 with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "InternScience/Agents-K1"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "InternScience/Agents-K1",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/InternScience/Agents-K1

SGLang

How to use InternScience/Agents-K1 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 "InternScience/Agents-K1" \
    --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": "InternScience/Agents-K1",
		"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 "InternScience/Agents-K1" \
        --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": "InternScience/Agents-K1",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use InternScience/Agents-K1 with Docker Model Runner:
```
docker model run hf.co/InternScience/Agents-K1
```

Leon-AI commited on 16 days ago

Commit

53cc997

verified ·

1 Parent(s): 1a11a5e

Update README.md

Browse files

Files changed (1) hide show

README.md +0 -16

README.md CHANGED Viewed

@@ -123,22 +123,6 @@ the model from defaulting to non-empty outputs).
   - JSON validity: `0.1 · 𝟙[valid JSON dict]` (or `0.05` for non-dict valid JSON)
   - task F1: `0.7 · F1(pred, gold)` — entity-set F1 for NER, triple-set F1 for RE
-### Hyper-parameters
-| | |
-|---|---|
-| Base model | Qwen3-4B-Instruct |
-| Learning rate | 1e-6 |
-| KL loss coeff. (low-var KL) | 0.01 |
-| Train batch size (prompts) | 128 |
-| PPO mini / micro batch | 64 / 4 per GPU |
-| Rollouts per prompt (`n`) | 6 |
-| Sampling | T=0.7, top-p=0.9 |
-| Max prompt / response | 2048 / 2048 |
-| Epochs | 1 (112 optimizer steps) |
-| Rollout engine | vLLM (gpu_memory_utilization=0.6) |
-| Distributed | FSDP via Ray, 8 × H200 GPUs |
-| Wall-clock | ~1 h 18 min |
 ## Evaluation

   - JSON validity: `0.1 · 𝟙[valid JSON dict]` (or `0.05` for non-dict valid JSON)
   - task F1: `0.7 · F1(pred, gold)` — entity-set F1 for NER, triple-set F1 for RE
 ## Evaluation