A
Alpha
completed

super_chat_20260306122507_hu7r

88.33M parameter nanochat model — bpe-chat-4k tokenizer, 16L/512D/8H

Full run detailsOpen in Chat

Overview

88.33M
Parameters
6.8672
Final Loss
6.9053
Best Val Loss
960.2
Perplexity
1,024,000
Tokens Processed
0.0
Tokens/Param
597 tok/s
Avg Throughput
5m 25s
Training Time
Training Progress500 / 500 steps (100.0%)
Loss reduced by 18.2% from initial 8.3976

Dataset & Training

Domainnanochat
Tokenizerbpe-chat-4k
Total Iterations500
Batch Size4
Context Length512 tokens
Tokens per Batch2,048
Dataset Passes~3
Effective Tokens1,024,000

Training Pipeline

Warmupsteps 1–5

Learning rate warmup — model weights adjusting to data distribution

Loss: 8.398 → 8.285Linear LR warmup, gradient clipping

Training Metrics

Loss Curve
?
?
?
?
Smoothed Loss
Perplexity
Learning Rate
Gradient Norm
Throughput (tok/s)
Timing Breakdown
No Telemetry

Model Architecture

Model Configuration

ArchitectureGPT (decoder-only transformer)
Parameters88.33M
Layers16
Embedding Dim512
Attention Heads8
Head Dim64
FFN Dim2048
FFN Activationswiglu
Vocab Size4,000
Context Length512 tokens
Dropout0

Training Configuration

Optimizeradamw
Learning Rate0.0006
LR Min0.00006
LR ScheduleCosine decay
Warmup Steps500
Batch Size4
Grad Accum Steps1
Effective Batch4
Grad Clip1
Weight Decay0.1
Backendhelios
Tokenizerbpe-chat-4k
Seed42

Layer Structure

Token Embed
4,000×512
Pos Embed
512×512
Block 0
Attn+FFN
Block 1
Attn+FFN
Block 2
Attn+FFN
Block 3
Attn+FFN
Block 4
Attn+FFN
Block 5
Attn+FFN
...10 more
LayerNorm
512
LM Head
512×4,000

Generated Samples

No samples generated yet. Samples appear at configured intervals during training.

Checkpoints

StepFileSizeDate
500checkpoint-500.json303.4 MBMar 6, 2026 1:05 PM

Chat with Model

Send a message to chat with this model
Generated Invalid Date Invalid Date — Alpha Training SystemConfig hash: