ENGPT - v0

Our model, fine-tuned with Llama 3.1 8B, has been trained using a dataset generated from the EPDK and TEDAŞ regulations within the Turkish Electricity Distribution system.

Model Details

It can answer questions related to regulations and provide detailed information about specific materials.

Model Description

Developed by: hdogrukan ,ecokumus
Language(s) (NLP): Turkish
Finetuned from model : Llama 3.1 8B

Model Sources

Metrics

-https://wandb.ai/hdogrukan/Fine-tune%20llama-3.1-8b-it%20on%20Turkish%20Energy%20Sector%20V0/reports/Engpt-v0--Vmlldzo5MjIzMTIx

Uses

!pip install --upgrade transformers

import transformers
import torch

model_id = "hdogrukan/Llama-3.1-8B-Instruct-Energy"

pipeline = transformers.pipeline(
    "text-generation",
    model=model_id,
    model_kwargs={"torch_dtype": torch.bfloat16},
    device_map="auto",
)


messages = [
    {"role": "system", "content": "You are helpful asistant."},
    {"role": "user", "content": "TEDAŞ-MLZ/99-032.E şartnamesini hangi kurum yayınlamıştır?"},
]

outputs = pipeline(
    messages,
    max_new_tokens=1024,
    temperature=0.2
)
# For more accuracy try this!
""" 
outputs = pipeline(
    messages,
    max_new_tokens=512,
    #max_length=50,
    num_return_sequences=3,
    do_sample=True,
    top_k=50,
    top_p=0.95,
    temperature=0.5 )
"""
print(outputs[0]["generated_text"][-1])

Output : "TEDAŞ-MLZ/99-032.E şartnamesini Türkiye Elektrik Dağıtım A.Ş. yayınlamıştır."