mistral-ivf-08

Model Trained Using AutoTrain

This model was trained using AutoTrain. For more information, please visit AutoTrain.

Usage

from transformers import AutoModelForCausalLM, AutoTokenizer

model_path = "PATH_TO_THIS_REPO"

tokenizer = AutoTokenizer.from_pretrained(model_path)
model = AutoModelForCausalLM.from_pretrained(
    model_path,
    device_map="auto",
    torch_dtype='auto'
).eval()

# Prompt content: "hi"
messages = [
    {"role": "user", "content": "hi"}
]

input_ids = tokenizer.apply_chat_template(conversation=messages, tokenize=True, add_generation_prompt=True, return_tensors='pt')
output_ids = model.generate(input_ids.to('cuda'))
response = tokenizer.decode(output_ids[0][input_ids.shape[1]:], skip_special_tokens=True)

# Model response: "Hello! How can I assist you today?"
print(response)

Extracting and Transforming the Training Dataset

To prepare the training dataset:

import pandas as pd

# Load the dataset
file_path = 'Dataset.csv'
df = pd.read_csv(file_path)

# Select the first 60,000 rows
subset_df = df.head(60000)

# Rename the column to 'text' for training compatibility
subset_df.rename(columns={"Conversational Format": "text"}, inplace=True)

# Save the subset to a new CSV file named train.csv
subset_df.to_csv('train.csv', index=False)

print("A new dataset with 60,000 rows has been saved as 'train.csv'.")

After cleaning and transformation, the dataset was reduced to 63,162 rows and three columns:

• text: containing the conversational examples.
• concept: summarizing the prediction task.
• description: explaining the dataset's purpose.