EzEpoch Training Dashboard

📊 Live Metrics

Loss

—

Current training loss

Learning Rate

—

Current LR

GPU Util

—

Compute utilization

GPU Memory

—

VRAM usage

GPU Temp

—

Temperature °C

GPU Power

—

Watts drawn

Epoch

—

Current / Total

Batch Size

—

BS × GA = effective

Steps / sec

—

Training throughput

Tokens / sec

—

Estimated tok/s

MFU

—

Model FLOPs util

ETA

—

Time remaining

Cost

—

Estimated GPU cost

Progress

—

Overall completion

Disk

—

Last Checkpoint

—

Most recent save

🤖 AI Guardian & Analysis

🎯 Mode:

LATEST ANALYSIS

AI analysis will appear here once training data is available…

ACTIVITY LOG

Loading activity…

💬 Ask Training AI

🎮 Training Controls

⚙️ GPU Memory & Training Settings

GPU Memory Target: 80%

70% Safe80% Balanced90% Aggressive

Lower = safer (avoids OOM) · Higher = faster (larger effective batch)

📈 Training Charts

Training Metrics

Training Loss

Loss Stability (Rolling Std Dev)

Learning Rate

GPU Memory

GPU Memory (GB)

GPU Memory %

GPU Memory Breakdown (Model · Grad · Optim · Activ · Cache · Free)

GPU Memory Timeline

GPU Health

GPU Utilization %

GPU Temperature

GPU Power Draw

Performance & Efficiency

Steps / sec

Tokens / sec

MFU %

Cost Tracker ($)

Advanced Analysis

🔬 Training Scattergram (interactive)

vs colored by

Points colored from ■ early → ■ mid → ■ late in training

Gradient Norm

Loss vs Learning Rate

ETA Remaining (hours)

Multi-GPU Balance (Utilization %)

Checkpoint Timeline

🧹 Data Quality Metrics

Dataset Size

—

Training samples

Avg Sample Len

—

Tokens per sample

Max Seq Length

—

Context window

Total Tokens

—

In training set

Data Quality

—

Overall score

📜 Training Log

[--:--:--]Waiting for training to start…

💾 Checkpoints

Select a session to view checkpoints…

📦 Model Downloads

⏳ Checking status…

Preparing… 0%