Instructions to use rtharrin/Qwen2.5-3B-ProxyMath-Merged with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use rtharrin/Qwen2.5-3B-ProxyMath-Merged with Transformers:

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

pipe = pipeline("text-generation", model="rtharrin/Qwen2.5-3B-ProxyMath-Merged")

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("rtharrin/Qwen2.5-3B-ProxyMath-Merged")
model = AutoModelForCausalLM.from_pretrained("rtharrin/Qwen2.5-3B-ProxyMath-Merged", device_map="auto")

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use rtharrin/Qwen2.5-3B-ProxyMath-Merged with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "rtharrin/Qwen2.5-3B-ProxyMath-Merged"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "rtharrin/Qwen2.5-3B-ProxyMath-Merged",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Use Docker

docker model run hf.co/rtharrin/Qwen2.5-3B-ProxyMath-Merged

SGLang

How to use rtharrin/Qwen2.5-3B-ProxyMath-Merged 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 "rtharrin/Qwen2.5-3B-ProxyMath-Merged" \
    --host 0.0.0.0 \
    --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "rtharrin/Qwen2.5-3B-ProxyMath-Merged",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

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 "rtharrin/Qwen2.5-3B-ProxyMath-Merged" \
        --host 0.0.0.0 \
        --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "rtharrin/Qwen2.5-3B-ProxyMath-Merged",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Docker Model Runner
How to use rtharrin/Qwen2.5-3B-ProxyMath-Merged with Docker Model Runner:
```
docker model run hf.co/rtharrin/Qwen2.5-3B-ProxyMath-Merged
```

rtharrin commited on 15 days ago

Commit

6c4d7fa

verified ·

1 Parent(s): c306459

Update README.md

Browse files

Files changed (1) hide show

README.md +1 -1

README.md CHANGED Viewed

@@ -21,7 +21,7 @@ Low-Rank Adaptation (LoRA) was selected over full fine-tuning to mitigate high c
 ## 4. Evaluation
 Our evaluation covers four target domains across arithmetic chain-of-thought, reading comprehension, factual alignment, and broad multi-disciplinary science domains. GSM8K CoT tracks raw problem-solving capability and was our target benchmark to increase; MMLU provides a calibrated measure of foundational knowledge retention across STEM and social domains; RACE assesses general contextual synthesis; and TruthfulQA quantifies post-tuning hallucinatory trends, as many fine tuning methods are known to create overconfidence. To establish a validated architectural baseline, we benchmark against the original instruction-tuned base model alongside two comparison setups of equal 3B scale: an identical Qwen iteration trained over a minimal 2% hyper-curated slice of direct task-specific data (340 rows of gsm8k), and an un-tuned baseline variant. Relative to these comparison targets, the proxy-trained model demonstrates superior structural alignment by boosting strict target math match constraints by 75%.
 | Benchmark Task / Split | Our Model (Numina-Proxy) |  Curated Baseline (2% GSM8K) | Base Qwen2.5-3B-Instruct (Baseline) |
-| :--- | :---: | :---: | :---: | :---: |
 | **GSM8K CoT (Strict EM)** | **0.0599** | 0.0281 | 0.0341 |
 | **GSM8K CoT (Flexible Extract)** | 0.5254 | 0.5921 | **0.5974** |
 | **MMLU (Overall Acc)** | 0.5264 | **0.6463** | **0.6463** |

 ## 4. Evaluation
 Our evaluation covers four target domains across arithmetic chain-of-thought, reading comprehension, factual alignment, and broad multi-disciplinary science domains. GSM8K CoT tracks raw problem-solving capability and was our target benchmark to increase; MMLU provides a calibrated measure of foundational knowledge retention across STEM and social domains; RACE assesses general contextual synthesis; and TruthfulQA quantifies post-tuning hallucinatory trends, as many fine tuning methods are known to create overconfidence. To establish a validated architectural baseline, we benchmark against the original instruction-tuned base model alongside two comparison setups of equal 3B scale: an identical Qwen iteration trained over a minimal 2% hyper-curated slice of direct task-specific data (340 rows of gsm8k), and an un-tuned baseline variant. Relative to these comparison targets, the proxy-trained model demonstrates superior structural alignment by boosting strict target math match constraints by 75%.
 | Benchmark Task / Split | Our Model (Numina-Proxy) |  Curated Baseline (2% GSM8K) | Base Qwen2.5-3B-Instruct (Baseline) |
+| :--- | :---: | :---: | :---: |
 | **GSM8K CoT (Strict EM)** | **0.0599** | 0.0281 | 0.0341 |
 | **GSM8K CoT (Flexible Extract)** | 0.5254 | 0.5921 | **0.5974** |
 | **MMLU (Overall Acc)** | 0.5264 | **0.6463** | **0.6463** |