Instructions to use mostafaamiri/persian_llama_7B_merged with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use mostafaamiri/persian_llama_7B_merged with Transformers:

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

pipe = pipeline("text-generation", model="mostafaamiri/persian_llama_7B_merged")

# pip install -U transformers accelerate
# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("mostafaamiri/persian_llama_7B_merged")
model = AutoModelForCausalLM.from_pretrained("mostafaamiri/persian_llama_7B_merged", device_map="auto")

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use mostafaamiri/persian_llama_7B_merged with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "mostafaamiri/persian_llama_7B_merged"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "mostafaamiri/persian_llama_7B_merged",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Use Docker

docker model run hf.co/mostafaamiri/persian_llama_7B_merged

SGLang

How to use mostafaamiri/persian_llama_7B_merged 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 "mostafaamiri/persian_llama_7B_merged" \
    --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": "mostafaamiri/persian_llama_7B_merged",
		"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 "mostafaamiri/persian_llama_7B_merged" \
        --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": "mostafaamiri/persian_llama_7B_merged",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Docker Model Runner
How to use mostafaamiri/persian_llama_7B_merged with Docker Model Runner:
```
docker model run hf.co/mostafaamiri/persian_llama_7B_merged
```

Questions for training

by alibidaran - opened Jan 1, 2024

Discussion

alibidaran

Jan 1, 2024

Did you train your tokenizer for Persian corpus?
How many train steps do yo preserve for training llama?

mostafaamiri

Owner Jan 3, 2024

Yes, the steps taken to train the model include:
1- Tokenizer training on persina corpus
2- Training LoRA model adapter on persian corpus
3- Instruct tuning on translated ALPACA and some similar corpus

alibidaran

Jan 3, 2024

I have the ALPACA Persian-style dataset, but I don't know how many training steps are required for training LLAMA2. I trained LLAMA2 for various English datasets, but Farsi even with a trained tokenizer, doesn't give me a considerable result.

mostafaamiri

Owner Jan 4, 2024

Do you train adapter on Farsi dataset.
I trained it on 200 milions token.

mostafaamiri changed discussion status to closed Apr 6, 2024

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