Instructions to use cpayne1303/cp2024-instruct with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use cpayne1303/cp2024-instruct with Transformers:

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

pipe = pipeline("text-generation", model="cpayne1303/cp2024-instruct")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("cpayne1303/cp2024-instruct")
model = AutoModelForCausalLM.from_pretrained("cpayne1303/cp2024-instruct")
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 cpayne1303/cp2024-instruct with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "cpayne1303/cp2024-instruct"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "cpayne1303/cp2024-instruct",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/cpayne1303/cp2024-instruct

SGLang

How to use cpayne1303/cp2024-instruct 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 "cpayne1303/cp2024-instruct" \
    --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": "cpayne1303/cp2024-instruct",
		"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 "cpayne1303/cp2024-instruct" \
        --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": "cpayne1303/cp2024-instruct",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use cpayne1303/cp2024-instruct with Docker Model Runner:
```
docker model run hf.co/cpayne1303/cp2024-instruct
```

Model Description

This is a model using the llama2 architecture and only 30 million parameters. It is based off of this model and was finetuned on approximately 85 million tokens of instruct data from the first 20000 rows of the openhermes 2.5 dataset with a low learning rate of 2e-6 and context length of 512.

Open LLM Leaderboard Evaluation Results

Detailed results can be found here

Metric	Value
Avg.	4.32
IFEval (0-Shot)	17.06
BBH (3-Shot)	2.48
MATH Lvl 5 (4-Shot)	0.00
GPQA (0-shot)	1.34
MuSR (0-shot)	3.18
MMLU-PRO (5-shot)	1.85

Downloads last month: 6

Safetensors

Model size

31.2M params

Tensor type

BF16

Model tree for cpayne1303/cp2024-instruct

Base model

cpayne1303/cp2024

Finetuned

(1)

this model

Dataset used to train cpayne1303/cp2024-instruct

Evaluation results

strict accuracy on IFEval (0-Shot)
Open LLM Leaderboard

17.060
normalized accuracy on BBH (3-Shot)
Open LLM Leaderboard

2.480
exact match on MATH Lvl 5 (4-Shot)
Open LLM Leaderboard

0.000
acc_norm on GPQA (0-shot)
Open LLM Leaderboard

1.340
acc_norm on MuSR (0-shot)
Open LLM Leaderboard

3.180
accuracy on MMLU-PRO (5-shot)
test set Open LLM Leaderboard

1.850