File size: 2,995 Bytes
64382c6
 
 
 
 
37b0b2a
64382c6
 
 
37b0b2a
64382c6
 
37b0b2a
 
 
64382c6
 
37b0b2a
64382c6
 
37b0b2a
64382c6
 
 
 
 
 
 
 
 
 
37b0b2a
64382c6
 
37b0b2a
64382c6
 
 
 
 
 
 
 
 
 
 
 
37b0b2a
64382c6
 
 
37b0b2a
64382c6
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
37b0b2a
64382c6
 
 
37b0b2a
 
 
 
 
 
 
 
 
 
 
 
 
 
64382c6
37b0b2a
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
import torch
from datasets import load_dataset
from unsloth import FastLanguageModel, UnslothTrainer, unsloth_train

# Load dataset
file_path = "/content/debug_divas_dataset.json"  # Ensure the file path is correct
dataset = load_dataset("json", data_files=file_path)

# Load Unsloth's FastLanguageModel and tokenizer
model_name = "unsloth/mistral-7b-instruct"  # Ensure it's an instruct model for translation
model, tokenizer = FastLanguageModel.from_pretrained(
    model_name=model_name,
    max_seq_length=128,
    dtype=torch.float32,  # Use float32 to avoid FP16 issues
    load_in_4bit=False,   # Disable 4-bit quantization if not needed
)

# Preprocessing function
def preprocess_function(examples):
    inputs = tokenizer(
        [f"Translate the following English sentence to colloquial Tamil: {text}" for text in examples["input"]],
        padding="max_length",
        truncation=True,
        max_length=128,
    )
    labels = tokenizer(
        examples["output"], padding="max_length", truncation=True, max_length=128
    )
    inputs["labels"] = labels["input_ids"]
    return inputs

# Tokenize dataset
tokenized_datasets = dataset.map(preprocess_function, batched=True, remove_columns=dataset["train"].column_names)

# Split dataset
split_datasets = tokenized_datasets["train"].train_test_split(test_size=0.2, seed=42)
train_dataset, test_dataset = split_datasets["train"], split_datasets["test"]

# Initialize UnslothTrainer
trainer = UnslothTrainer(
    model=model,
    train_dataset=train_dataset,
    eval_dataset=test_dataset,
    tokenizer=tokenizer,
    args={
        "per_device_train_batch_size": 8,
        "per_device_eval_batch_size": 8,
        "num_train_epochs": 3,
        "learning_rate": 2e-5,
        "save_strategy": "epoch",
        "evaluation_strategy": "epoch",
        "fp16": False,  # Disable mixed precision training
    }
)

# Train with Unsloth
unsloth_train(trainer)

# Save fine-tuned model
trainer.model.save_pretrained("./fine_tuned_model")
tokenizer.save_pretrained("./fine_tuned_model")

# Load fine-tuned model
fine_tuned_model, tokenizer = FastLanguageModel.from_pretrained(
    model_name="./fine_tuned_model",
    max_seq_length=128,
    dtype=torch.float32,
    load_in_4bit=False,
)

# Move model to device
device = "cuda" if torch.cuda.is_available() else "cpu"
fine_tuned_model.to(device)

# User input loop for real-time translation
print("Colloquial Tamil Translator (Type 'exit' to quit)")
while True:
    input_text = input("Enter an English sentence: ")
    if input_text.lower() == "exit":
        break
    
    instruction = "Translate the following English sentence to colloquial Tamil"
    
    inputs = tokenizer(f"{instruction}: {input_text}", return_tensors="pt").to(device)
    
    # Generate translation
    translated_tokens = fine_tuned_model.generate(**inputs)
    translated_text = tokenizer.decode(translated_tokens[0], skip_special_tokens=True)

    print("Colloquial Tamil Translation:", translated_text)