Instructions to use Cannae-AI/ReasoningLlama-Math-1B-IT with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use Cannae-AI/ReasoningLlama-Math-1B-IT with Transformers:

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

pipe = pipeline("text-generation", model="Cannae-AI/ReasoningLlama-Math-1B-IT")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("Cannae-AI/ReasoningLlama-Math-1B-IT")
model = AutoModelForCausalLM.from_pretrained("Cannae-AI/ReasoningLlama-Math-1B-IT")
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 Cannae-AI/ReasoningLlama-Math-1B-IT with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "Cannae-AI/ReasoningLlama-Math-1B-IT"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "Cannae-AI/ReasoningLlama-Math-1B-IT",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/Cannae-AI/ReasoningLlama-Math-1B-IT

SGLang

How to use Cannae-AI/ReasoningLlama-Math-1B-IT 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 "Cannae-AI/ReasoningLlama-Math-1B-IT" \
    --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": "Cannae-AI/ReasoningLlama-Math-1B-IT",
		"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 "Cannae-AI/ReasoningLlama-Math-1B-IT" \
        --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": "Cannae-AI/ReasoningLlama-Math-1B-IT",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use Cannae-AI/ReasoningLlama-Math-1B-IT with Docker Model Runner:
```
docker model run hf.co/Cannae-AI/ReasoningLlama-Math-1B-IT
```

CannaeAI commited on Nov 18, 2025

Commit

5867878

verified ·

1 Parent(s): 3031856

Update README.md

Browse files

Files changed (1) hide show

README.md +3 -14

README.md CHANGED Viewed

@@ -1,6 +1,6 @@
 ---
 base_model:
-- unsloth/Llama-3.2-1B
 tags:
 - text-generation-inference
 - transformers
@@ -19,18 +19,7 @@ library_name: transformers
 ---
 # ReasoningLlama-Math-1B-IT
 ## Model Description:
-This is a fine-tuned version of [unsloth/Llama-3.2-1B](https://huggingface.co/unsloth/Llama-3.2-1B) on the [unsloth/OpenMathReasoning-mini](https://huggingface.co/datasets/unsloth/OpenMathReasoning-mini) dataset which is a small version of the [nvidia/OpenMathReasoning](https://huggingface.co/datasets/nvidia/OpenMathReasoning) dataset which was used to win the [AIMO](https://www.kaggle.com/competitions/ai-mathematical-olympiad-progress-prize-2/leaderboard) (AI Mathematical Olympiad) challenge!
 - **recommended settings for inference:** min_p = 0.1 and temperature = 1.5 , Read this [Tweet](https://x.com/menhguin/status/1826132708508213629) to understand why.
 - **License :** apache-2.0
-- **Finetuned from model :** unsloth/Llama-3.2-1B
-## Benchmarks:
-We evaluate ReasoningLlama-Math-1B-IT on GSM8K using the standard lm-eval 5-shot exact-match protocol. Under identical decoding and extraction settings,ReasoningLlama-Math-1B-IT outperforms Meta’s Llama-3.2-1B by more than 4×, demonstrating a substantial improvement in small-model mathematical reasoning capability.
-| Model                         | Params | GSM8K (5-shot, EM) |
-| ----------------------------- | ------ | ------------------ |
-| **ReasoningLlama-Math-1B-IT** | 1B     | **30.7%**          |
-| Llama-3.2-1B                  | 1B     | 6.8%               |
-<p align="center">
-  <img alt="ReasonningLlama-Math-1B-IT" src="https://huggingface.co/Cannae-AI/ReasoningLlama-Math-1B-IT/resolve/main/ChatGPT%20Image%2018%20nov.%202025%2C%2020_55_23.png">
-</p>

 ---
 base_model:
+- unsloth/Llama-3.2-1B-Instruct
 tags:
 - text-generation-inference
 - transformers
 ---
 # ReasoningLlama-Math-1B-IT
 ## Model Description:
+This is a fine-tuned version of [unsloth/Llama-3.2-1B-Instruct](https://huggingface.co/unsloth/Llama-3.2-1B-Instruct) on the [unsloth/OpenMathReasoning-mini](https://huggingface.co/datasets/unsloth/OpenMathReasoning-mini) dataset which is a small version of the [nvidia/OpenMathReasoning](https://huggingface.co/datasets/nvidia/OpenMathReasoning) dataset which was used to win the [AIMO](https://www.kaggle.com/competitions/ai-mathematical-olympiad-progress-prize-2/leaderboard) (AI Mathematical Olympiad) challenge!
 - **recommended settings for inference:** min_p = 0.1 and temperature = 1.5 , Read this [Tweet](https://x.com/menhguin/status/1826132708508213629) to understand why.
 - **License :** apache-2.0
+- **Finetuned from model :** unsloth/Llama-3.2-1B-Instruct