Instructions to use Phind/Phind-CodeLlama-34B-Python-v1 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use Phind/Phind-CodeLlama-34B-Python-v1 with Transformers:

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

pipe = pipeline("text-generation", model="Phind/Phind-CodeLlama-34B-Python-v1")

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("Phind/Phind-CodeLlama-34B-Python-v1")
model = AutoModelForCausalLM.from_pretrained("Phind/Phind-CodeLlama-34B-Python-v1")

Notebooks
Google Colab
Kaggle
Local Apps

vLLM

How to use Phind/Phind-CodeLlama-34B-Python-v1 with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "Phind/Phind-CodeLlama-34B-Python-v1"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "Phind/Phind-CodeLlama-34B-Python-v1",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Use Docker

docker model run hf.co/Phind/Phind-CodeLlama-34B-Python-v1

SGLang

How to use Phind/Phind-CodeLlama-34B-Python-v1 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 "Phind/Phind-CodeLlama-34B-Python-v1" \
    --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": "Phind/Phind-CodeLlama-34B-Python-v1",
		"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 "Phind/Phind-CodeLlama-34B-Python-v1" \
        --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": "Phind/Phind-CodeLlama-34B-Python-v1",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Docker Model Runner
How to use Phind/Phind-CodeLlama-34B-Python-v1 with Docker Model Runner:
```
docker model run hf.co/Phind/Phind-CodeLlama-34B-Python-v1
```

This might be the best coding model (Better than wizardcoder_python_34b)

by rombodawg - opened Aug 27, 2023

Discussion

rombodawg

Aug 27, 2023

•

edited Aug 27, 2023

This is the first model to pass my "code snake in python" test. The only other model that passed this was gpt 3.5

And yes i tested wizardcoder_python_34b it failed the code snake in python test

The following prompt was used to create the code for snake in text generation web ui

Prompt:
I need you to write 1 piece of python code for a functioning game of snake. Make sure to fill every line of code completely, do not leave any place holders, as logic for the code must be thought of now and implemented in this generation.

Make sure the game have a window to play in, has proper working food for the snake to collect, has the ability for the snake to eat the food and grow from the eaten food, has the ability for the snake to die from running into itself or the border of the window, and has a restart button once the game is over to play the game over again.

Make sure the game has a simple visual graphical interface for the user to see when playing, you will have to code this yourself. Absolutely make sure everything mentioned in the code is written, do not leave a class that is mentioned somewhere but not written out properly elsewhere Please follow this format when writing the code:
FILENAME

CODE

CuriousRick

Aug 27, 2023

Thanks for sharing the Template, have you tweaked parameters like temperature or top-p to run it?, if so, can you share those?

rombodawg

Aug 27, 2023

I ran it on oobagooba web ui with the default simple-1 preset

acrastt

Aug 27, 2023

This is the first model to pass my "code snake in python" test. The only other model that passed this was gpt 3.5

And yes i tested wizardcoder_python_34b it failed the code snake in python test

The following prompt was used to create the code for snake in text generation web ui

Prompt:
I need you to write 1 piece of python code for a functioning game of snake. Make sure to fill every line of code completely, do not leave any place holders, as logic for the code must be thought of now and implemented in this generation.

Make sure the game have a window to play in, has proper working food for the snake to collect, has the ability for the snake to eat the food and grow from the eaten food, has the ability for the snake to die from running into itself or the border of the window, and has a restart button once the game is over to play the game over again.

Make sure the game has a simple visual graphical interface for the user to see when playing, you will have to code this yourself. Absolutely make sure everything mentioned in the code is written, do not leave a class that is mentioned somewhere but not written out properly elsewhere Please follow this format when writing the code:
FILENAME
CODE

Have you tested without giving it a lecture about how games work? No one would have time to write all of that in real-world scenarios. Also, you said only GPT-3.5 got this correct, GPT-4 doesnt? Claude 2?

michaelroyzen

Phind org Aug 28, 2023

Stay tuned -- instruction-tuned V2 models are coming tomorrow :)

rombodawg

Aug 28, 2023

@michaelroyzen Please give us 13b tuned models too 🙏🙏🙏

Upload images, audio, and videos by dragging in the text input, pasting, or clicking here.

Tap or paste here to upload images

· Sign up or log in to comment