A
Alpha
20260224053053_zek5completedchat6.84M params54m 27s elapsed · Updated 48d ago
6L / 256D / 8H · helios · bpe-4k · adamw· Created Feb 24, 2026 5:31 AM
Step 5,836 / 50,00011.7%
5.7909
Loss?
5.3617
Best Loss?
-30.7% from start
6.0736
Val Loss?
best: 5.9736
2.92e-5
Learning Rate?
7,573
Throughput?
tok/s (avg)
541
Speed?
ms/iter (avg)
4292.889
Grad Norm?
avg: 6971.933
23.67M
Tokens
processed
63ms
Forward
12% of step
416ms
Backward
77% of step
25ms
GPU Sync
5% of step
1,064
GPU Ops
per step
1.0%
MFU
model FLOPS util
6.6x
Bwd/Fwd
ratio
Loss Curve ? click any chart to add markers
?
?
?
?
Architecture
Layers?6
Embedding?256
Heads?8
Vocab?4,000
Context?256
Dropout?0.1
Parameters?6.84M
Training Config
Total iters?50,000
Batch size?16
Max LR?0.00003
Optimizer?adamw
Backend?helios
Tokenizer?bpe-4k
Seed?42
Weight decay?0.1
Grad clip?1
Eval interval?2500
Throughput (tok/s)
Step Time (ms/iter)
GPU & VRAM
Perplexity
Train/Val Gap
Learning Rate
Grad Norm
Smoothed Loss (EMA)
Loss Velocity
Gradient Clipping
No clipping data
GPU Operations
Step Time Breakdown
Forward
Backward
Grad Norm
Optimizer
GPU Sync
Data
Timing Phase Lines
Backward / Forward Ratio
Checkpoints (0) ?
No checkpoints saved
Sample Generations (3)
#CheckpointPrompt (preview)Generated
1-<|user|> Hello, how are you? <|assistant|>48d ago
Prompt
<|user|> Hello, how are you? <|assistant|>
Output
<|user|> Hello, how are you? <|assistant|>er—that in the cof our sed without erarc—not the ned by must the unnyesas of the fto s as may cds from I an—the true st; awars we an,
2-<|user|> What do you like to do for fun? <|assistant|>48d ago
Prompt
<|user|> What do you like to do for fun? <|assistant|>
Output
<|user|> What do you like to do for fun? <|assistant|>; your n; your acin both to act to both sing the a ging or its of the cred. <|assistant|> If in human is to are your acy t to a of the pening to the cit and it the to no not
3-<|user|> Tell me about yourself. <|assistant|>48d ago
Prompt
<|user|> Tell me about yourself. <|assistant|>
Output
<|user|> Tell me about yourself. <|assistant|>s and the and s y a for a in exin which to exors from which our tmay the moral in the eder and not unged we and to a moral y y, to as in a, to yal, ping p
Model Config (JSON)
{
  "vocabSize": 4000,
  "blockSize": 256,
  "nLayer": 6,
  "nEmbd": 256,
  "nHead": 8,
  "dropout": 0.1
}
Training Config (JSON)
{
  "iters": 50000,
  "batchSize": 16,
  "lr": 0.00003,
  "lrMin": 0.000003,
  "warmupIters": 500,
  "beta1": 0.9,
  "beta2": 0.99,
  "eps": 1e-8,
  "weightDecay": 0.1,
  "gradClip": 1,
  "evalInterval": 2500,
  "evalIters": 10,
  "seed": 42,
  "backend": "helios",
  "tokenizer": "bpe-4k",
  "optimizer": "adamw",
  "logLevel": "info",
  "trace": false,
  "gradAccumSteps": 1,
  "sampleInterval": 500,
  "spikeThreshold": 0
}