Instructions to use NinedayWang/PolyCoder-2.7B with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use NinedayWang/PolyCoder-2.7B with Transformers:

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

pipe = pipeline("text-generation", model="NinedayWang/PolyCoder-2.7B")

# Load model directly
from transformers import AutoTokenizer, AutoModelForMultimodalLM

tokenizer = AutoTokenizer.from_pretrained("NinedayWang/PolyCoder-2.7B")
model = AutoModelForMultimodalLM.from_pretrained("NinedayWang/PolyCoder-2.7B")

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use NinedayWang/PolyCoder-2.7B with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "NinedayWang/PolyCoder-2.7B"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "NinedayWang/PolyCoder-2.7B",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Use Docker

docker model run hf.co/NinedayWang/PolyCoder-2.7B

SGLang

How to use NinedayWang/PolyCoder-2.7B 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 "NinedayWang/PolyCoder-2.7B" \
    --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": "NinedayWang/PolyCoder-2.7B",
		"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 "NinedayWang/PolyCoder-2.7B" \
        --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": "NinedayWang/PolyCoder-2.7B",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Docker Model Runner
How to use NinedayWang/PolyCoder-2.7B with Docker Model Runner:
```
docker model run hf.co/NinedayWang/PolyCoder-2.7B
```

About to help with this model to application maintenance

by joluinfante - opened Jun 9, 2024

Discussion

joluinfante

Jun 9, 2024

I'm trying to simplify the maintenance of a c/c++ application from a number of sources, assuming that Polycoder is already trained in c and c++, by re-training it with my application logic. I understand that it is not simply "make the sources available and let the AI do magic", but, as I read, I also need to inform you about the relationship between all functions (function names, input and output parameters) using AST or similar. Since that will require a lot of processing power, I want to try a simple program, with a few functions. Obviously, I know little about this, I'm trying to learn. I understand that I have to use the tokenizer to adapt this information to what is required by the model, and then generate the model. And, finally, through a prompt, try to establish what things I can ask him to respond to, based on my training. I mean, I don't want magic. Just an iterative process where you learn how to train, and how to ask. Could you give me a simple example of how to start with this, in python? That is, I have seen that you can be asked to produce code, but starting from general training. I intend to re-train (or expand knowledge) from a simple application.

joluinfante

Jun 9, 2024

Continue reading about this. It would seem that I have to create a dataset with my example source code. But, I don't want to make a new dataset (because if so, I would have to teach it about c and c++). I want to add knowledge to an existing dataset (in this case, the Polycoder one). Or simply by loading many sources in c and c++ would I be producing a dataset that allows the AI to learn to code in c and c++ from my source code?

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