🚀 Qwen3-30M TinyStories Pretrained (FP16) - Notebook Version

Pretrained Qwen3-30M model on TinyStories dataset using FP16 precision in notebook environment.

📊 Training Results

Final Training Loss: 1.5244
Final Validation Loss: 1.5601832866668701
Training Samples: -1
Epochs: 3
Precision: FP16
Dataset: TinyStories (child-friendly stories)

🚀 Usage

from transformers import AutoTokenizer, AutoModelForCausalLM
import torch

tokenizer = AutoTokenizer.from_pretrained("Mostafa8Mehrabi/qwen3-30m-tinystories-final")
model = AutoModelForCausalLM.from_pretrained(
    "Mostafa8Mehrabi/qwen3-30m-tinystories-final", 
    torch_dtype=torch.float16,
    device_map="auto"
)

# Generate a story
prompt = "Once upon a time, there was a little girl named"
inputs = tokenizer(prompt, return_tensors="pt")
outputs = model.generate(**inputs, max_length=200, do_sample=True, temperature=0.7)
print(tokenizer.decode(outputs[0], skip_special_tokens=True))

📁 Checkpoints

Training checkpoints (also in FP16) are available at: Mostafa8Mehrabi/qwen3-30m-tinystories-checkpoints

📖 About TinyStories Dataset

The TinyStories dataset contains simple, child-friendly stories that are perfect for:

Story generation
Child-safe content creation
Educational applications
Creative writing assistance

🔧 Training Environment

This model was trained in a notebook environment with the following configuration:

Batch Size: 128
Learning Rate: 5e-05
Max Length: 512
Number of Processes: 8