A
Alpha
historic_chat_v2_20260225123612_2ao8activechat17.44M params34m 37s elapsed · ~36h 42m remaining
8L / 384D / 8H · helios · bpe-4k · adamw· Created Feb 25, 2026 12:36 PM
Step 860 / 50,0001.7%
6.4525
Loss?
6.4517
Best Loss?
-22.9% from start
-
Val Loss?
4.37e-5
Learning Rate?
2,287
Throughput?
tok/s (avg)
2,689
Speed?
ms/iter (avg)
0.851
Grad Norm?
avg: 7.151
5.22M
Tokens
processed
243ms
Forward
9% of step
2398ms
Backward
89% of step
14ms
GPU Sync
1% of step
782
GPU Ops
per step
0.8%
MFU
model FLOPS util
9.9x
Bwd/Fwd
ratio
Loss Curve ?
Symbio semantics: this chart stitches many fresh candidate evaluations onto one global step axis. Loss resets near switches are expected because candidates are re-initialized. Compare local candidate shapes and the global frontier, not a single continuous model trajectory.
Search semantics
Validation / selection
Run diagnostics
Search Trajectory + Frontier
Candidate-local train/val loss segments on a shared step axis, with switch events and global frontier overlays.
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
Clip Telemetry
SymbiogenesisSWIGLU
1.76
Wt Entropy
bits
20.0
Eff. Rank
6.4700
Free Energy
3.911
Pop Entropy
nats
0.0740
Complexity
0.0602
Fitness
449
CUSUM Alerts
of 850 steps
8
Batch Size
adaptive
CUSUM Change-Point Monitor
Weight Entropy
Effective Rank
Free Energy
Fitness Score
Population Entropy
Adaptive Batch Size
Phase Change / Gelation
Current
Transitioning
Stability
41%
Phase Changes
10
Regime Shifts
0
Training dynamics are shifting. The model may be entering a new loss basin or the learning rate is hitting a critical threshold. This often happens before a breakthrough or a plateau.
Phase Timeline
Step 1Step 811
Loss Oscillation (Harmonic Analysis)
Evolutionary Search
Generations
2
Candidates
8
Activations
4
Best Loss
8.1824
Total Steps
160
#CandidateActivationGenLossFitnessStepsMutation
1gen1_gelu_7gelu18.18240.033620origin
2gen1_silu_8silu18.21830.034020origin
3gen1_gelu_6gelu18.23800.033620origin
4gen0_relu_3relu08.25750.032720origin
5gen1_gelu_5gelu18.26820.032020origin
6gen0_silu_2silu08.28560.032820origin
7gen0_gelu_1gelu08.31440.031720origin
8gen0_swiglu_4swiglu08.33240.030320origin
Generation Summary
G04c8.2575
G14c8.1824
Fitness Progression
Architecture Diversity
Convergence vs Diversity (Tug-of-War)
Current Mode
Converging
Diversity Pressure
52%
Convergence Momentum
100%
Convergence Progress
100%
Phase Portrait: Diversity Pressure vs Convergence Momentum
Low diversity / high momentum = lock-in convergence
High diversity / high momentum = productive exploration
Low diversity / low momentum = stalled collapse
High diversity / low momentum = diversity stalling convergence
Tug-of-War Trace (Time Domain)
Positive tension means recent frontier improvement is outpacing diversity pressure (search is converging). Negative tension means exploration pressure is dominating recent convergence momentum (search is broadening or getting “stumped”).
Strongest Convergence
step 81
tension 0.480
Strongest Diversity Push
step 1
tension -0.800
Best Frontier
8.1824
progress 100%
Evolutionary Lineage Tree
Lineage Tree
100%
Activation Flow (Sankey)
Activation Switch Log
StepFromToGenPrev StepsBest LossFinal LossFitnessTree
1-gelu0----
21gelusilu0208.31448.31440.0317
41silurelu0208.28568.28560.0328
61reluswiglu0208.25758.26700.0327
81swiglugelu1208.33248.33240.0303
101gelugelu1208.26828.26820.0320
121gelugelu1208.23808.23800.0336
141gelusilu1208.18248.18240.0336
Search Candidates
#NameActivationGenParentStepsBest LossBest ValAvg LossFitnessAvg tok/sAlerts
1gen1_gelu_7gelu1-208.1824-8.26770.03367,20320
2gen1_silu_8silu1-208.2183-8.30090.03402,32320
3gen1_gelu_6gelu1-208.2380-8.30530.03367,19020
4gen0_relu_3relu0-208.2575-8.31450.03277,17520
5gen1_gelu_5gelu1-208.2682-8.32410.03207,18720
6gen0_silu_2silu0-208.2856-8.33570.03282,30120
7gen0_gelu_1gelu0-208.3144-8.34660.03177,21310
8gen0_swiglu_4swiglu0-208.3324-8.36050.03032,21320
Activation Distribution
gelu
80 (50%)
silu
40 (25%)
relu
20 (13%)
swiglu
20 (13%)
Oscillation & Heat Capacity
Activation Evolution Radial
Symbio Config
{
  "cusumSensitivity": 4,
  "cusumBaselineWindow": 5,
  "metricsInterval": 10,
  "trackWeightEntropy": true,
  "trackEffectiveRank": true,
  "trackFreeEnergy": true,
  "trackMIProfiles": true,
  "trackPopulationMetrics": true,
  "freeEnergyBeta": 0.01,
  "miNumBins": 30,
  "adaptiveBatch": true,
  "batchMin": 8,
  "batchMax": 64,
  "batchStep": 4,
  "calmStepsBeforeRestore": 200,
  "fitnessAlpha": 1,
  "complexityMode": "entropy",
  "diversityBonus": 0,
  "diversityDecay": "none",
  "searchMode": "ffn-activation-search",
  "activationPool": [
    "gelu",
    "silu",
    "relu",
    "swiglu"
  ],
  "searchStrategy": "evolutionary",
  "populationSize": 4,
  "generations": 2,
  "selectionStrategy": "topk",
  "tournamentK": 3,
  "mutationRate": 0.5,
  "stepsPerCandidate": 20,
  "rankBy": "valLoss",
  "perfWeight": 0,
  "stabilityWeight": 0,
  "writeReport": true,
  "writeCandidates": true,
  "writeSummary": true
}
Checkpoints (0) ?
No checkpoints saved
Sample Generations (3)
#CheckpointPrompt (preview)Generated
1-<|user|> Hello, how are you? <|assistant|>10h ago
Prompt
<|user|> Hello, how are you? <|assistant|>
Output
<|user|> Hello, how are you? <|assistant|>ing a diis it through both s and of to I ues y; s for a this b’s to rey and the claughter in’s s, as a gods and to with a la, for must thusly I ced? <|assistant|>
2-<|user|> What do you like to do for fun? <|assistant|>10h ago
Prompt
<|user|> What do you like to do for fun? <|assistant|>
Output
<|user|> What do you like to do for fun? <|assistant|>to to for and of the cenent a as the ces our pto that as with a fto enesnot might ts a of the a to a ced by est , er or the cs, in a menor in unes from s, and
3-<|user|> Tell me about yourself. <|assistant|>10h ago
Prompt
<|user|> Tell me about yourself. <|assistant|>
Output
<|user|> Tell me about yourself. <|assistant|>sining it that inaed and true the sdeάDies for an pyour pass the smay not and invirtue ing our our of the diging accountI a I dient by that into to
{
  "vocabSize": 4000,
  "blockSize": 512,
  "nLayer": 8,
  "nEmbd": 384,
  "nHead": 8,
  "dropout": 0.1,
  "ffnActivation": "swiglu",
  "ffnDim": 1024
}
{
  "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": 10,
  "syncEvery": 1,
  "gcEvery": 0,
  "packed": false,
  "symbio": true,
  "symbioConfig": {
    "cusumSensitivity": 4,
    "cusumBaselineWindow": 5,
    "metricsInterval": 10,
    "trackWeightEntropy": true,
    "trackEffectiveRank": true,
    "trackFreeEnergy": true,
    "trackMIProfiles": true,
    "trackPopulationMetrics": true,
    "freeEnergyBeta": 0.01,
    "miNumBins": 30,
    "adaptiveBatch": true,
    "batchMin": 8,
    "batchMax": 64,
    "batchStep": 4,
    "calmStepsBeforeRestore": 200,
    "fitnessAlpha": 1,
    "complexityMode": "entropy",
    "diversityBonus": 0,
    "diversityDecay": "none",
    "searchMode": "ffn-activation-search",
    "activationPool": [
      "gelu",
      "silu",
      "relu",
      "swiglu"
    ],
    "searchStrategy": "evolutionary",
    "populationSize": 4,
    "generations": 2,
    "selectionStrategy": "topk",
    "tournamentK": 3,
    "mutationRate": 0.5,
    "stepsPerCandidate": 20,
    "rankBy": "valLoss",
    "perfWeight": 0,
    "stabilityWeight": 0,
    "writeReport": true,
    "writeCandidates": true,
    "writeSummary": true
  }
}