README.md

metadata

datasets:
  - tatsu-lab/alpaca
license: other
license_name: ncsam
license_link: https://github.com/FlameF0X/NCSAM
language:
  - en
base_model:
  - FlameF0X/SnowflakeCore-G1-Tiny
pipeline_tag: text-generation
library_name: transformers
tags:
  - pre_train
  - costume_code

SnowflakeCore-G1-Tiny-Instruct

A custom GPT-style transformer language model built from scratch using PyTorch.

Model Overview

SnowflakeCore-G1-Tiny and SnowflakeCore-G1-Tiny-Instruct are a GPT-style autoregressive transformer model with ~407M parameters (407,334,912) designed for text generation tasks.

Key Features

• 2048 token context window for extended conversations
• Mixed precision training (BF16/FP16) for efficiency
• Custom attention implementation with fused operations
• Early stopping mechanisms N/A
• Gradient accumulation for effective large batch training

Architecture Specifications

Component	Value
Model Type	Autoregressive Transformer
Parameters	~407M
Layers	24
Hidden Size	1024
Attention Heads	16
Head Dimension	64
FFN Dimension	4096
Context Length	2048 tokens
Vocabulary Size	50,257 (GPT-2 tokenizer)

Quick Start

Installation

pip install torch transformers # if not already installed

Basic Usage

# N/A

Training Details

Dataset

• Source:

Training Configuration

• Framework: PyTorch with mixed precision (BF16/FP16)
• Optimizer: AdamW (learning rate: 2e-4)
• Batch Size: N/A
• Context Window: 2048 tokens or 512 tokens
• Validation Split: N/A
• Early Stopping: N/A

Performance Monitoring

• Training loss tracked per epoch with perplexity calculation
• Full validation after each epoch
• Step-level monitoring every 500 steps
• Comprehensive metrics saved in training_metrics.json

Technical Implementation

Attention Mechanism

• Causal Masking: Supports autoregressive generation
• Key Padding Mask: Enables batched inference
• Scaled Dot-Product: Head dimension normalization included

Memory Optimization

• Fused Operations: Reduces memory fragmentation
• Mixed Precision: 30-40% memory reduction
• Gradient Accumulation: Simulates larger batch sizes
• Optional Quantization: Further model compression

Training Stability

• Gradient Clipping: Prevents exploding gradients
• Automatic Loss Scaling: Mixed precision stability
• Early Stopping: Prevents overfitting with patience mechanisms

System Requirements

Memory Requirements

• Training: 16-24GB VRAM (precision dependent)
• Inference: 4-6GB VRAM for standard generation
• Context: Maximum 2048 tokens input length

Generation Parameters

Default configuration:

{
  "do_sample": true,
  "temperature": 1.0,
  "top_p": 0.9,
  "top_k": 50,
  "max_new_tokens": 50,
  "pad_token_id": 50256,
  "eos_token_id": 50256
}

Limitations

• No HuggingFace .generate() support: Use custom generation function
• Output Quality: May produce repetitive or nonsensical text for some prompts
• Hardware Requirements: GPU recommended for practical inference
• Context Window: Limited to 2048 tokens (or 512 tokens)

Example Output

# WIP

License & Acknowledgments

• License: NCSAM
• Framework: Built using PyTorch
• Dataset:

Support Me

You can support me via Ko-fi or you can try my Vast.ai template!

More meta-data

• Release date: July 10 2025