Instructions to use issai/foggen-qwen3-1.7b with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use issai/foggen-qwen3-1.7b with Transformers:

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

pipe = pipeline("text-generation", model="issai/foggen-qwen3-1.7b")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("issai/foggen-qwen3-1.7b")
model = AutoModelForCausalLM.from_pretrained("issai/foggen-qwen3-1.7b")
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 issai/foggen-qwen3-1.7b with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "issai/foggen-qwen3-1.7b"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "issai/foggen-qwen3-1.7b",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/issai/foggen-qwen3-1.7b

SGLang

How to use issai/foggen-qwen3-1.7b 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 "issai/foggen-qwen3-1.7b" \
    --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": "issai/foggen-qwen3-1.7b",
		"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 "issai/foggen-qwen3-1.7b" \
        --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": "issai/foggen-qwen3-1.7b",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use issai/foggen-qwen3-1.7b with Docker Model Runner:
```
docker model run hf.co/issai/foggen-qwen3-1.7b
```
Browse Quantizations to use this model in llama.cpp, Ollama, LM Studio, or any compatible app.

FogGen (Qwen3-1.7B): cross-scale R14 endpoint

The 1.7B-parameter scale-up of issai/foggen. Same 14-round self-evolving recipe, same cloud teacher, same seven MCQ domains; the edge backbone is the only change.

This checkpoint exists to test whether the FogGen recipe transfers across model size. The canonical 0.6B endpoint lives at issai/foggen and is the deployment model. This 1.7B variant trades cost for raw and system accuracy: same recipe, larger backbone, ~3× the per-query compute.

For background on the system overview, training pipeline, and routing protocol, see the issai/foggen model card; only the differences are documented here.

Recipe

Everything is held identical to issai/foggen:

Cloud teacher: Qwen3-30B-A3B-Instruct-2507
7 domain rotation: finance, science, coding, law, math, Kazakh culture, medical
14 sequential SFT rounds (R0 → R14)
LoRA r=16, α=32, all-linear, bf16, 2 epochs, lr=5e-5
Same confidence buckets (0.0, 0.25, 0.5, 0.75, 1.0) and same FogGen output format
For R0 the 1,800-question calibration buffer is re-labeled from scratch with the raw Qwen3-1.7B base (N=8 at T=0.7)

The only change is the edge backbone (Qwen/Qwen3-1.7B in place of Qwen/Qwen3-0.6B).

Performance

System accuracy at τ=0.5 on the seven MCQ domains (full test sets, ~16,200 queries). Cloud baseline is Qwen3-30B-A3B-Instruct-2507.

Domain	Cloud only	R14 raw	Random @ τ=0.5	FogGen @ τ=0.5	Cloud routed
Finance	69.5%	66.0%	66.7%	71.0%	20.1%
Science	72.7%	64.0%	66.1%	70.7%	23.2%
Coding	74.2%	68.9%	69.9%	73.7%	18.3%
Law	70.7%	62.1%	64.7%	69.5%	29.8%
Math	60.1%	47.2%	51.8%	56.7%	35.6%
Kazakh culture	95.8%	90.7%	91.1%	92.5%	8.2%
Medical	74.0%	62.0%	65.7%	72.1%	30.9%
Mean	73.9%	65.8%	68.0%	72.3%	23.7%

Mean lift over Random at τ=0.5: +4.3.

Quick demo

from transformers import AutoTokenizer, AutoModelForCausalLM

model = AutoModelForCausalLM.from_pretrained("issai/foggen-qwen3-1.7b", torch_dtype="bfloat16", device_map="auto")
tokenizer = AutoTokenizer.from_pretrained("issai/foggen-qwen3-1.7b")

SYSTEM = """You are a self-aware multiple-choice assistant.

Rules:
- Do not output <think> tags.
- First, assess your confidence in solving this question.
- Then give your answer.
- Output format:
  Confidence: <0.0|0.25|0.5|0.75|1.0>
  Final answer: <OPTION_LETTER>"""

messages = [
    {"role": "system", "content": SYSTEM},
    {"role": "user", "content": "<your MCQ here>"},
]
inputs = tokenizer.apply_chat_template(messages, return_tensors="pt", add_generation_prompt=True,
                                       enable_thinking=False).to(model.device)
outputs = model.generate(inputs, max_new_tokens=64, do_sample=False)
print(tokenizer.decode(outputs[0][inputs.shape[-1]:], skip_special_tokens=True))

The routing decision (route_query helper, threshold τ) is identical to the issai/foggen card.

Comparison to issai/foggen

	`issai/foggen` (0.6B)	`issai/foggen-qwen3-1.7b` (this)
Edge backbone	Qwen3-0.6B	Qwen3-1.7B
Per-query inference cost	1×	~3×
Mean R14 raw acc	59.6%	65.8%
Mean system acc @ τ=0.5	67.8%	72.3%
Cloud-routing rate @ τ=0.5	21.9%	23.7%