Instructions to use Norod78/Hebrew-GPT2-345M-Stage with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use Norod78/Hebrew-GPT2-345M-Stage with Transformers:

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

pipe = pipeline("text-generation", model="Norod78/Hebrew-GPT2-345M-Stage")

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("Norod78/Hebrew-GPT2-345M-Stage")
model = AutoModelForCausalLM.from_pretrained("Norod78/Hebrew-GPT2-345M-Stage")

llama-cpp-python

How to use Norod78/Hebrew-GPT2-345M-Stage with llama-cpp-python:

# !pip install llama-cpp-python

from llama_cpp import Llama

llm = Llama.from_pretrained(
	repo_id="Norod78/Hebrew-GPT2-345M-Stage",
	filename="ggml-model-f16.gguf",
)

output = llm(
	"Once upon a time,",
	max_tokens=512,
	echo=True
)
print(output)

Notebooks
Google Colab
Kaggle
Local Apps Settings

llama.cpp

How to use Norod78/Hebrew-GPT2-345M-Stage with llama.cpp:

Install from brew

brew install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama-server -hf Norod78/Hebrew-GPT2-345M-Stage:F16
# Run inference directly in the terminal:
llama-cli -hf Norod78/Hebrew-GPT2-345M-Stage:F16

Install from WinGet (Windows)

winget install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama-server -hf Norod78/Hebrew-GPT2-345M-Stage:F16
# Run inference directly in the terminal:
llama-cli -hf Norod78/Hebrew-GPT2-345M-Stage:F16

Use pre-built binary

# Download pre-built binary from:
# https://github.com/ggerganov/llama.cpp/releases
# Start a local OpenAI-compatible server with a web UI:
./llama-server -hf Norod78/Hebrew-GPT2-345M-Stage:F16
# Run inference directly in the terminal:
./llama-cli -hf Norod78/Hebrew-GPT2-345M-Stage:F16

Build from source code

git clone https://github.com/ggerganov/llama.cpp.git
cd llama.cpp
cmake -B build
cmake --build build -j --target llama-server llama-cli
# Start a local OpenAI-compatible server with a web UI:
./build/bin/llama-server -hf Norod78/Hebrew-GPT2-345M-Stage:F16
# Run inference directly in the terminal:
./build/bin/llama-cli -hf Norod78/Hebrew-GPT2-345M-Stage:F16

Use Docker

docker model run hf.co/Norod78/Hebrew-GPT2-345M-Stage:F16

LM Studio
Jan

vLLM

How to use Norod78/Hebrew-GPT2-345M-Stage with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "Norod78/Hebrew-GPT2-345M-Stage"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "Norod78/Hebrew-GPT2-345M-Stage",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Use Docker

docker model run hf.co/Norod78/Hebrew-GPT2-345M-Stage:F16

SGLang

How to use Norod78/Hebrew-GPT2-345M-Stage 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 "Norod78/Hebrew-GPT2-345M-Stage" \
    --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": "Norod78/Hebrew-GPT2-345M-Stage",
		"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 "Norod78/Hebrew-GPT2-345M-Stage" \
        --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": "Norod78/Hebrew-GPT2-345M-Stage",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Ollama
How to use Norod78/Hebrew-GPT2-345M-Stage with Ollama:
```
ollama run hf.co/Norod78/Hebrew-GPT2-345M-Stage:F16
```

Unsloth Studio

How to use Norod78/Hebrew-GPT2-345M-Stage with Unsloth Studio:

Install Unsloth Studio (macOS, Linux, WSL)

curl -fsSL https://unsloth.ai/install.sh | sh
# Run unsloth studio
unsloth studio -H 0.0.0.0 -p 8888
# Then open http://localhost:8888 in your browser
# Search for Norod78/Hebrew-GPT2-345M-Stage to start chatting

Install Unsloth Studio (Windows)

irm https://unsloth.ai/install.ps1 | iex
# Run unsloth studio
unsloth studio -H 0.0.0.0 -p 8888
# Then open http://localhost:8888 in your browser
# Search for Norod78/Hebrew-GPT2-345M-Stage to start chatting

Using HuggingFace Spaces for Unsloth

# No setup required
# Open https://huggingface.co/spaces/unsloth/studio in your browser
# Search for Norod78/Hebrew-GPT2-345M-Stage to start chatting

Docker Model Runner
How to use Norod78/Hebrew-GPT2-345M-Stage with Docker Model Runner:
```
docker model run hf.co/Norod78/Hebrew-GPT2-345M-Stage:F16
```

Lemonade

How to use Norod78/Hebrew-GPT2-345M-Stage with Lemonade:

Pull the model

# Download Lemonade from https://lemonade-server.ai/
lemonade pull Norod78/Hebrew-GPT2-345M-Stage:F16

Run and chat with the model

lemonade run user.Hebrew-GPT2-345M-Stage-F16

List all available models

lemonade list

Hebrew-GPT2-345M-Stage

An undertrained GPT2 based Hebrew text generation model which I slightly trained at 2020 on text from "Bama Hadasha" ("במה חדשה") A gguf version is available here

Dataset

Around 10% of the text from stage.co.il

LM Studio

A configuration scheme for LM Studio is available here

Usage with Transformers - sample code

import os
os.environ["PYTORCH_ENABLE_MPS_FALLBACK"] = "1"

from transformers import pipeline, set_seed
import random

model_id = "Norod78/Hebrew-GPT2-345M-Stage"
text_generator = pipeline('text-generation', model=model_id, tokenizer=model_id, device_map="auto")
max_length = 256
top_k = 70
top_p = 0.92
temperature = 1.0
max_seed = (2**32)-1
global_seed = random.randint(0, max_seed)

def text_generation(input_text = ''):
    global global_seed
    global_seed = global_seed + 1
    if global_seed >= max_seed:
        global_seed = 0
    if input_text == None or len(input_text) == 0:
        input_text = "\n"
    set_seed(global_seed)
    generated_text = text_generator(input_text,
    max_length=max_length,
    top_k=top_k, 
    top_p=top_p,    
    temperature=temperature,
    do_sample=True,
    repetition_penalty=1.4,
    num_return_sequences=1)
    parsed_text = generated_text[0]["generated_text"].replace("<|startoftext|>", "").replace("\r","").replace("\n\n", "\n").replace("\t", " ").replace("<|pad|>", " * ").replace("\"\"", "\"").strip()
    #print("parsed_text = \"" + parsed_text + "\" (seed = " + str(global_seed) + ")")
    return parsed_text

def main():
    prompt_prefix = "\n\n שם היצירה: "
    prompt_text = prompt_prefix + "חגבים ירוקים מקפצים בשדה"
    result = text_generation(input_text=prompt_text)
    print("result : \n" + str(result))

if __name__ == '__main__':
    main()

Downloads last month: 826

Safetensors

Model size

0.4B params

Tensor type

F32

Norod78
/

Hebrew-GPT2-345M-Stage

Hebrew-GPT2-345M-Stage

Dataset

Around 10% of the text from stage.co.il

LM Studio

Usage with Transformers - sample code

Space using Norod78/Hebrew-GPT2-345M-Stage 1