Instructions to use 4bit/Replit-v1-CodeInstruct-3B with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use 4bit/Replit-v1-CodeInstruct-3B with Transformers:

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

pipe = pipeline("text-generation", model="4bit/Replit-v1-CodeInstruct-3B", trust_remote_code=True)

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("4bit/Replit-v1-CodeInstruct-3B", trust_remote_code=True)
model = AutoModelForCausalLM.from_pretrained("4bit/Replit-v1-CodeInstruct-3B", trust_remote_code=True)

Notebooks
Google Colab
Kaggle
Local Apps

vLLM

How to use 4bit/Replit-v1-CodeInstruct-3B with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "4bit/Replit-v1-CodeInstruct-3B"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "4bit/Replit-v1-CodeInstruct-3B",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Use Docker

docker model run hf.co/4bit/Replit-v1-CodeInstruct-3B

SGLang

How to use 4bit/Replit-v1-CodeInstruct-3B 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 "4bit/Replit-v1-CodeInstruct-3B" \
    --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": "4bit/Replit-v1-CodeInstruct-3B",
		"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 "4bit/Replit-v1-CodeInstruct-3B" \
        --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": "4bit/Replit-v1-CodeInstruct-3B",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Docker Model Runner
How to use 4bit/Replit-v1-CodeInstruct-3B with Docker Model Runner:
```
docker model run hf.co/4bit/Replit-v1-CodeInstruct-3B
```

Replit-v1-CodeInstruct-3B / README.md

camenduru

thanks to teknium ❤

8b543a0 almost 3 years ago

preview code

raw

history blame contribute delete

2.7 kB

	---
	license: cc-by-sa-4.0
	datasets:
	- bigcode/the-stack-dedup
	- sahil2801/CodeAlpaca-20k
	- teknium/GPTeacher-CodeInstruct
	model-base:
	- replit/replit-code-v1-3b
	tags:
	- code
	- instruct
	- self instruct
	language:
	- code
	programming_language:
	- Markdown
	- Java
	- JavaScript
	- Python
	- TypeScript
	- PHP
	- SQL
	- JSX
	- reStructuredText
	- Rust
	- C
	- CSS
	- Go
	- C++
	- HTML
	- Vue
	- Ruby
	- Jupyter Notebook
	- R
	- Shell
	---

	Base Model: replit/replit-code-v1-3b

	This model is fine tuned on both Sahil2801's CodeAlpaca & Teknium's GPTeacher Code-Instruct to give Replit's Code model instruct capabilities.

	Try this model on it's HuggingFace demo Spaces: https://huggingface.co/spaces/teknium/Replit-v1-CodeInstruct-3B

	Dataset links:
	CodeAlpaca: https://huggingface.co/datasets/sahil2801/CodeAlpaca-20k
	GPTeacher subset - Code Instruct: https://github.com/teknium1/GPTeacher

	This model was trained on 2x a100 80gb for 1 hour on ~25,000 code instruction/response pairs in Alpaca format.

	Refer to the base models HuggingFace model card for some basic requirements to run: https://huggingface.co/replit/replit-code-v1-3b

	This fine tune can be prompted like any alpaca fine tune:
	```
	### Instruction:
	<prompt>

	### Input:
	<additional context>

	### Response:
	```

	or

	```
	### Instruction:
	<prompt>

	### Response:

	```

	This model seems to have issues with device="auto" in the model arguments (and requires the trust_remote_code=True, so you should maybe load it like I am here:
	```
	self.tokenizer = AutoTokenizer.from_pretrained("./Replit-CodeInstruct/", trust_remote_code=True)
	self.model = AutoModelForCausalLM.from_pretrained(
	"./Replit-CodeInstruct",
	torch_dtype=torch.bfloat16,
	trust_remote_code=True
	)
	self.model.to('cuda')
	```


	This model for me produced coherent outputs with the following sampler settings, but feel free to experiment:
	```
	max_new_tokens=128, do_sample=True, use_cache=True, temperature=0.2, top_p=0.9, eos_token_id= self.tokenizer.eos_token_id
	```

	In the tokenizer decode arguments, it also needs these settings:
	```
	skip_special_tokens=True, clean_up_tokenization_space=False
	```

	The following parameters were used with HuggingFace trainer to train the model with:
	```
	--model_name_or_path replit/replit-code-v1-3b --data_path /root/stanford_alpaca/train.json --bf16 True --output_dir /root/stanford_alpaca/model_ckpts --num_train_epochs 3 --per_device_train_batch_size 4 --per_device_eval_batch_size 1 --gradient_accumulation_steps 8 --save_strategy steps --save_steps 200 --save_total_limit 3 --learning_rate 1e-5 --weight_decay 0. --warmup_ratio 0.03 --tf32 True --run_name Replit1
	```