A
Alpha
super_chat_20260225082546_kyc4activechat17.44M params55m 33s elapsed · ~12h 49m remaining
8L / 384D / 8H · helios · bpe-4k · adamw· Created Feb 25, 2026 8:26 AM
Step 3,419 / 50,0006.8%
4.4502
Loss?
4.0052
Best Loss?
-46.8% from start
4.1977
Val Loss?
best: 4.1977
4.97e-5
Learning Rate?
6,203
Throughput?
tok/s (avg)
991
Speed?
ms/iter (avg)
1.186
Grad Norm?
avg: 1.180
21.01M
Tokens
processed
211ms
Forward
21% of step
712ms
Backward
72% of step
10ms
GPU Sync
1% of step
830
GPU Ops
per step
2.1%
MFU
model FLOPS util
3.4x
Bwd/Fwd
ratio
Loss Curve ?
Validation / selection
Run diagnostics
Gradient + Loss
Train/val loss with validation, warmup, overfit, and instability markers.
Search-aware view
Architecture
Layers?8
Embedding?384
Heads?8
Vocab?4,000
Context?512
Dropout?0.1
Parameters?17.44M
Training Config
Total iters?50,000
Batch size?12
Max LR?0.00005
Optimizer?adamw
Backend?helios
Tokenizer?bpe-4k
Seed?42
Weight decay?0.1
Grad clip?1
Eval interval?1000
GPU & VRAM
Learning Rate
Grad Norm
Step Time Breakdown
Step Time Breakdown
Checkpoints (0) ?
No checkpoints saved
Sample Generations (3)
#CheckpointPrompt (preview)Generated
1-<|user|> Hello, how are you? <|assistant|>15h ago
Prompt
<|user|> Hello, how are you? <|assistant|>
Output
<|user|> Hello, how are you? <|assistant|>aret <|user|> S Ls a <|assistant|> T <|assistant|> mm<|end_of_text|> <|user|> I <|assistant|> I’S <|assistant|> Y the ing m<|assistant|> I <|assistant|> not v TIit's w<|user|> No <|user|> U rka <|assistant|> cgor
2-<|user|> What do you like to do for fun? <|assistant|>15h ago
Prompt
<|user|> What do you like to do for fun? <|assistant|>
Output
<|user|> What do you like to do for fun? <|assistant|>at the a, your a <|assistant|> An<|end_of_text|> <|user|> dpasn p, that <|assistant|> <|user|> I mv you <|assistant|> rnt <|assistant|> I <|user|> N b<|user|> i <|assistant|> OTaing <|assistant|> <|assistant|> I’s
3-<|user|> Tell me about yourself. <|assistant|>15h ago
Prompt
<|user|> Tell me about yourself. <|assistant|>
Output
<|user|> Tell me about yourself. <|assistant|>in <|assistant|> I dcu <|assistant|> the py<|end_of_text|> <|user|> the kre’t <|assistant|> OO<|assistant|> W <|user|> I’s the cahanas’s <|user|> Oo<|assistant|> L <|assistant|> T <|assistant|> a<|assistant|> I <|assistant|> 
{
  "vocabSize": 4000,
  "blockSize": 512,
  "nLayer": 8,
  "nEmbd": 384,
  "nHead": 8,
  "dropout": 0.1
}
{
  "iters": 50000,
  "batchSize": 12,
  "lr": 0.00005,
  "lrMin": 0.000005,
  "warmupIters": 1000,
  "beta1": 0.9,
  "beta2": 0.95,
  "eps": 0.000001,
  "weightDecay": 0.1,
  "gradClip": 1,
  "evalInterval": 1000,
  "evalIters": 10,
  "seed": 42,
  "backend": "helios",
  "tokenizer": "bpe-4k",
  "optimizer": "adamw",
  "logLevel": "info",
  "trace": false,
  "gradAccumSteps": 1,
  "sampleInterval": 500,
  "spikeThreshold": 0,
  "syncEvery": 1,
  "gcEvery": 0,
  "packed": false
}