Instructions to use monsterapi/zephyr-7b-alpha_metamathqa with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use monsterapi/zephyr-7b-alpha_metamathqa with Transformers:

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

pipe = pipeline("text-generation", model="monsterapi/zephyr-7b-alpha_metamathqa")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("monsterapi/zephyr-7b-alpha_metamathqa")
model = AutoModelForCausalLM.from_pretrained("monsterapi/zephyr-7b-alpha_metamathqa")
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 monsterapi/zephyr-7b-alpha_metamathqa with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "monsterapi/zephyr-7b-alpha_metamathqa"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "monsterapi/zephyr-7b-alpha_metamathqa",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/monsterapi/zephyr-7b-alpha_metamathqa

SGLang

How to use monsterapi/zephyr-7b-alpha_metamathqa 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 "monsterapi/zephyr-7b-alpha_metamathqa" \
    --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": "monsterapi/zephyr-7b-alpha_metamathqa",
		"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 "monsterapi/zephyr-7b-alpha_metamathqa" \
        --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": "monsterapi/zephyr-7b-alpha_metamathqa",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use monsterapi/zephyr-7b-alpha_metamathqa with Docker Model Runner:
```
docker model run hf.co/monsterapi/zephyr-7b-alpha_metamathqa
```

Finetuning Overview:

Model Used: HuggingFaceH4/zephyr-7b-alpha
Dataset: meta-math/MetaMathQA

Dataset Insights:

The MetaMathQA dataset is a newly created dataset specifically designed for enhancing the mathematical reasoning capabilities of large language models (LLMs). It is built by bootstrapping mathematical questions and rewriting them from multiple perspectives, providing a comprehensive and challenging environment for LLMs to develop and refine their mathematical problem-solving skills.

Finetuning Details:

Using MonsterAPI's LLM finetuner, this finetuning:

Was conducted with efficiency and cost-effectiveness in mind.
Completed in a total duration of 10.9 hours for 0.5 epoch using an A6000 48GB GPU.
Costed $22.01 for the entire finetuning process.

Hyperparameters & Additional Details:

Epochs: 0.5
Total Finetuning Cost: $22.01
Model Path: HuggingFaceH4/zephyr-7b-alpha
Learning Rate: 0.0001
Data Split: 95% train 5% validation
Gradient Accumulation Steps: 4

Prompt Structure

Below is an instruction that describes a task. Write a response that appropriately completes the request.  


###Instruction:[query]


###Response:[response]

Training loss:

Benchmark Results:

GSM8K is a dataset of 8.5K high quality linguistically diverse grade school math word problems, These problems take between 2 and 8 steps to solve, and solutions primarily involve performing a sequence of elementary calculations using basic arithmetic operations (+ − ×÷) to reach the final answer. A bright middle school student should be able to solve every problem. Its a industry wide used benchmark for testing an LLM for for multi-step mathematical reasoning.

license: apache-2.0

Downloads last month: 25

Safetensors

Model size

7B params

Tensor type

BF16

Model tree for monsterapi/zephyr-7b-alpha_metamathqa

Base model

mistralai/Mistral-7B-v0.1

Finetuned

HuggingFaceH4/zephyr-7b-alpha

Finetuned

(14)

this model

monsterapi
/

zephyr-7b-alpha_metamathqa