Instructions to use interview-eval/zephyr-7b-math-train with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use interview-eval/zephyr-7b-math-train with Transformers:

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

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

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("interview-eval/zephyr-7b-math-train")
model = AutoModelForCausalLM.from_pretrained("interview-eval/zephyr-7b-math-train")
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 interview-eval/zephyr-7b-math-train with vLLM:

Install from pip and serve model

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

Use Docker

docker model run hf.co/interview-eval/zephyr-7b-math-train

SGLang

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

Docker Model Runner
How to use interview-eval/zephyr-7b-math-train with Docker Model Runner:
```
docker model run hf.co/interview-eval/zephyr-7b-math-train
```

zephyr-7b-math-train / README.md

Juyoung Suk

End of training

1f38b07 verified over 1 year ago

preview code

raw

history blame contribute delete

2.08 kB

	---
	library_name: transformers
	license: apache-2.0
	base_model: alignment-handbook/zephyr-7b-sft-full
	tags:
	- alignment-handbook
	- trl
	- sft
	- generated_from_trainer
	- trl
	- sft
	- generated_from_trainer
	datasets:
	- EunsuKim/MATH
	model-index:
	- name: zephyr-7b-math-train
	results: []
	---

	<!-- This model card has been generated automatically according to the information the Trainer had access to. You
	should probably proofread and complete it, then remove this comment. -->

	# zephyr-7b-math-train

	This model is a fine-tuned version of [alignment-handbook/zephyr-7b-sft-full](https://huggingface.co/alignment-handbook/zephyr-7b-sft-full) on the EunsuKim/MATH dataset.
	It achieves the following results on the evaluation set:
	- Loss: 0.0188

	## Model description

	More information needed

	## Intended uses & limitations

	More information needed

	## Training and evaluation data

	More information needed

	## Training procedure

	### Training hyperparameters

	The following hyperparameters were used during training:
	- learning_rate: 1e-05
	- train_batch_size: 16
	- eval_batch_size: 8
	- seed: 42
	- distributed_type: multi-GPU
	- num_devices: 4
	- total_train_batch_size: 64
	- total_eval_batch_size: 32
	- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
	- lr_scheduler_type: cosine
	- lr_scheduler_warmup_ratio: 0.1
	- num_epochs: 10

	### Training results

	\| Training Loss \| Epoch \| Step \| Validation Loss \|
	\|:-------------:\|:-----:\|:----:\|:---------------:\|
	\| 0.8757 \| 1.0 \| 5 \| 0.7950 \|
	\| 0.6949 \| 2.0 \| 10 \| 0.5316 \|
	\| 0.48 \| 3.0 \| 15 \| 0.3425 \|
	\| 0.2951 \| 4.0 \| 20 \| 0.1809 \|
	\| 0.1534 \| 5.0 \| 25 \| 0.0872 \|
	\| 0.0746 \| 6.0 \| 30 \| 0.0426 \|
	\| 0.0409 \| 7.0 \| 35 \| 0.0291 \|
	\| 0.0287 \| 8.0 \| 40 \| 0.0229 \|
	\| 0.022 \| 9.0 \| 45 \| 0.0196 \|
	\| 0.019 \| 10.0 \| 50 \| 0.0188 \|


	### Framework versions

	- Transformers 4.44.2
	- Pytorch 2.4.1+cu124
	- Datasets 2.21.0
	- Tokenizers 0.19.1