seta-sft-kimi-qwen3

Instructions to use AnonymousSubmissionUnderDouble-BlindRevi/seta-sft-kimi-qwen3 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use AnonymousSubmissionUnderDouble-BlindRevi/seta-sft-kimi-qwen3 with Transformers:

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

pipe = pipeline("text-generation", model="AnonymousSubmissionUnderDouble-BlindRevi/seta-sft-kimi-qwen3")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# pip install -U transformers accelerate
# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("AnonymousSubmissionUnderDouble-BlindRevi/seta-sft-kimi-qwen3")
model = AutoModelForCausalLM.from_pretrained("AnonymousSubmissionUnderDouble-BlindRevi/seta-sft-kimi-qwen3", device_map="auto")
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=256)
print(tokenizer.decode(outputs[0][inputs["input_ids"].shape[-1]:]))

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use AnonymousSubmissionUnderDouble-BlindRevi/seta-sft-kimi-qwen3 with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "AnonymousSubmissionUnderDouble-BlindRevi/seta-sft-kimi-qwen3"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "AnonymousSubmissionUnderDouble-BlindRevi/seta-sft-kimi-qwen3",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/AnonymousSubmissionUnderDouble-BlindRevi/seta-sft-kimi-qwen3

SGLang

How to use AnonymousSubmissionUnderDouble-BlindRevi/seta-sft-kimi-qwen3 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 "AnonymousSubmissionUnderDouble-BlindRevi/seta-sft-kimi-qwen3" \
    --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": "AnonymousSubmissionUnderDouble-BlindRevi/seta-sft-kimi-qwen3",
		"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 "AnonymousSubmissionUnderDouble-BlindRevi/seta-sft-kimi-qwen3" \
        --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": "AnonymousSubmissionUnderDouble-BlindRevi/seta-sft-kimi-qwen3",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use AnonymousSubmissionUnderDouble-BlindRevi/seta-sft-kimi-qwen3 with Docker Model Runner:
```
docker model run hf.co/AnonymousSubmissionUnderDouble-BlindRevi/seta-sft-kimi-qwen3
```

SETA-SFT: Qwen3-8B Fine-tuned with SFT on Kimi-K2.5 Trajectories

This model is Qwen3-8B fine-tuned with supervised fine-tuning (SFT) on trajectories collected from a strong teacher model (Kimi-K2.5) on SETA terminal-agent tasks.

It is a checkpoint released alongside the paper SETA: Scaling Environments for Terminal Agents (anonymous submission under double-blind review).

Model Details


Base model	Qwen/Qwen3-8B
Training method	Supervised Fine-Tuning (SFT)
Teacher model	Kimi-K2.5
Training data	1 488 terminal-agent trajectories collected on SETA tasks
Trainable tokens	~7.79 M (thinking variant)
Context length	32 768 tokens
Thinking	Enabled (trajectories preserve `<think>…</think>` reasoning blocks)

Intended Use

This model is designed for terminal agent tasks: completing multi-step shell-based tasks inside a Docker container environment using tools such as shell_exec, shell_view, and shell_write_content_to_file.

It serves as an SFT warm-start baseline and can be further improved with reinforcement learning.

Usage

from transformers import AutoModelForCausalLM, AutoTokenizer

model_id = "AnonymousSubmissionUnderDouble-BlindRevi/seta-sft-kimi-qwen3"
tokenizer = AutoTokenizer.from_pretrained(model_id)
model = AutoModelForCausalLM.from_pretrained(model_id, torch_dtype="auto", device_map="auto")

For evaluation with the SETA framework, serve via SGLang and run:

python -m sglang.launch_server --model AnonymousSubmissionUnderDouble-BlindRevi/seta-sft-kimi-qwen3 --port 30000

python scripts/evaluation/eval.py \
    --config scripts/evaluation/configs/eval_default_qwen3_8b.yaml \
    terminal_env.model.model_type=AnonymousSubmissionUnderDouble-BlindRevi/seta-sft-kimi-qwen3 \
    terminal_env.model.url=http://localhost:30000/v1 \
    dataset=seta-env

Training Configuration

Key hyperparameters (see scripts/areal_sft/configs/seta_kimi_qwen3_sft_thinking.yaml in the companion code repository):

Hyperparameter	Value
Learning rate	2e-5
LR scheduler	cosine
Optimizer	AdamW (β₁=0.9, β₂=0.95)
Weight decay	0.05
Total epochs	3
Batch size	8
Max sequence length	16 384 tokens
GPUs	2 × (d2p1t1)

SFT Data Summary

Trajectories were collected by running Kimi-K2.5 on SETA tasks and filtering for verified reward:

	Thinking variant
Total trajectories	1 488
Fully-passing trajectories	1 112
Mean reward	0.932
Total tokens	15.69 M
Trainable tokens	7.79 M (49.6%)

Limitations

Evaluated on terminal-agent benchmarks; performance on general language tasks is not characterized.
SFT trajectories were collected from a single teacher model; diversity may be limited.

License

Apache 2.0 (inherited from Qwen3-8B base).

Downloads last month: 15

Safetensors

Model size

8B params

Tensor type

BF16

Model tree for AnonymousSubmissionUnderDouble-BlindRevi/seta-sft-kimi-qwen3

Base model

Qwen/Qwen3-8B-Base

Finetuned

Qwen/Qwen3-8B

Finetuned

(2158)

this model