Instructions to use troysaved/claimtrace-smoke with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use troysaved/claimtrace-smoke with Transformers:

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

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

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("troysaved/claimtrace-smoke")
model = AutoModelForCausalLM.from_pretrained("troysaved/claimtrace-smoke", 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=40)
print(tokenizer.decode(outputs[0][inputs["input_ids"].shape[-1]:]))

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use troysaved/claimtrace-smoke with vLLM:

Install from pip and serve model

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

Use Docker

docker model run hf.co/troysaved/claimtrace-smoke

SGLang

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

Unsloth Desktop
Docker Model Runner
How to use troysaved/claimtrace-smoke with Docker Model Runner:
```
docker model run hf.co/troysaved/claimtrace-smoke
```

claimtrace-smoke / README.md

troysaved

card: mark as smoke-test artifact, point to the real submission

d1c7d44 verified 6 days ago

preview code

Raw

History Blame Contribute Delete

1.67 kB

	---
	base_model: unsloth/qwen3-1.7b-unsloth-bnb-4bit
	library_name: transformers
	license: apache-2.0
	language: [en]
	tags: [claimtrace, smoke-test, not-the-submission, qwen3, unsloth]
	---

	# claimtrace — smoke-test artifact (NOT the submission)

	> **Looking for the claimtrace model? It is
	> [`troysaved/claimtrace-qwen3-1.7b`](https://huggingface.co/troysaved/claimtrace-qwen3-1.7b).**
	> This repo is not that model and should not be evaluated as it.

	This is a throwaway checkpoint from a 6-conversation end-to-end pipeline test
	(`smoke.sh`: generate → train → eval on a tiny batch), uploaded on 2026-08-19 to
	prove the publish path worked before the real run was pushed. It was trained on a
	smoke batch, not on the dataset, and it holds no meaningful behavior. Its numbers
	are not reported anywhere and it is not referenced by the submission.

	It is kept public rather than deleted so that any link created while the publish
	path was being tested still resolves to an explanation instead of a 404.

	## The actual submission

	\| \| \|
	\|---\|---\|
	\| Model \| [`troysaved/claimtrace-qwen3-1.7b`](https://huggingface.co/troysaved/claimtrace-qwen3-1.7b) @ `d02f380b37e5ce2ea8d816ff8379718148874b66` (run `qadv1`) \|
	\| Dataset \| [`troysaved/claimtrace-ledger-dataset`](https://huggingface.co/datasets/troysaved/claimtrace-ledger-dataset) @ `9ee828685f5aba3d0f9ac4edf69a40b9b3dc82f9` (v4 spec, 407 conversations) \|
	\| Behavior \| A tutoring model that keeps a claim-provenance ledger: an item enters `KNOWN` only after the learner demonstrates it; a self-report is `CLAIMED`, however plausible. \|

	Trained with [Unsloth](https://github.com/unslothai/unsloth) and Hugging Face TRL.