Pre-launch· Token not live yet — no rewards or treasury until launchWhat this means
TRAIN · Solana memecoin intelligence

The memecoin training a model to understand memecoins.

TRAIN is our own model, trained from scratch on real Solana launches: their setup, their risks, and what happened to launches like them. Once the token launches, its creator rewards pay for the compute. Every dollar, run and benchmark is public.

Analyze a tokenTalk to TRAIN
TRAIN v0.19
Live
Current model
Decision-tree ensemble
Training treasury
$0.00
no token yet
Compute used
79.4 CPU min
CPU training · $0 so far
Training runs
25
Latest benchmark
0.955
Spotting crashes (−70% within 1h) · score 0–1, 0.5 = coin flip
Launches tracked
3,536
15,173 snapshots with observed outcomes

How TRAIN works

live values
  1. 01
    Token activity
    pump.fun
    trading generates creator fees
  2. 02
    Creator rewards
    $0.00
    no token yet
  3. 03
    Training treasury
    $0.00
    starts at token launch
  4. 04
    Compute
    79.4 CPU min
    CPU · $0 so far
  5. 05
    Training runs
    25
    0 failed
  6. 06
    Model versions
    25
    v0.19 live
  7. 07
    Benchmark
    0.955
    Spotting crashes (−70% within 1h)

Current model

Model details

TRAIN v0.19 is live. It reads each new launch's trading, holders, bundled and sniper wallets, creator history, fees paid and the track record of its buyers, and estimates what usually happens next: crashes, creator dumps, survival and 2× runs.

Spotting crashes (−70% within 1h)
0.955
Spotting survivors (still traded after 1h)
0.957
Spotting 2× runs (within 1h)
0.866
Calling the 1h direction (down / flat / up)
56.9%
Spotting creator dumps (within 1h)
0.880

Scores are measured on launches TRAIN never trained on. 1.0 is perfect, 0.5 is a coin flip.

Latest training

All runs
No training in progress. The next run starts automatically once enough new launches have observed outcomes.

Treasury

Ledger
$0.00
The TRAIN token hasn't launched yet, so there are no creator rewards and no treasury. Until launch, training is free and starts whenever enough new data has been collected.
Creator rewards received
$0.00
Spent on training
$0.00
Spent serving the model
$0.00

Model progress

Held-out test set v8
v0.010.450 – 0.961v0.25
Spotting crashes (−70% within 1h) per version. Dashed line = chance.
v0.25
0.944↑
v0.24
0.944↓
v0.23
0.947↓
v0.22
0.948↑
v0.21
0.924↓

Recent activity

Everything
  1. 00:45
    TRAIN v0.25 not deployed; v0.19 stays live
    Reason: average score across 5 tests 0.844 vs 0.845 for the live v0.19.
  2. 00:45
    Benchmark completed: TRAIN v0.25
    Spotting crashes (−70% within 1h) regressed 0.955 → 0.944 vs live v0.19.
  3. 00:45
    Benchmark started: TRAIN v0.25
  4. 00:45
    Training completed: TRAIN v0.25
    00:02:55 · trained on 9081 examples · $0 compute cost
  5. 00:42
    Training started: TRAIN v0.25
    Solana Launch Dataset v17 · decision-tree ensemble · TRAIN training server (CPU)
  6. 00:42
    Dataset updated: Solana Launch Dataset v17
    10,305 snapshots from 2,068 launches.
  7. 00:42
    TRAIN v0.25 queued
    Same configuration as v0.18 (best so far on the headline benchmark), retrained on the newest dataset.
  8. 00:42
    Enough new data for another training run
    6,368 new labelled snapshots since Solana Launch Dataset v9.