Pre-launch· Token not live yet — no rewards or treasury until launchWhat this means
← Model
Adds wallet track records

TRAIN v0.06

Adds the realized track record (inside TRAIN's sample, before the snapshot) of the wallets buying the launch — how many are known, profitable, losing or high-frequency.
Completed

Definition

Model type
Decision-tree ensemble
Signals used
Market + wallet history
Technical details
max_depth
3
subsample
0.8
n_estimators
300
learning_rate
0.03
min_samples_leaf
10
Random seed
1006
Config hash
44e154af79cc8325
Created
05 Oct 06:00
Completed
05 Oct 06:01
Artifact
models/v0.06/model.json
sha256 140a8d78cd40d8d9…

Dataset & compute

Dataset
Solana Launch Dataset v5
784 snapshots · 157 launches · sha256 a8050102d8fd
Training runs
1
Compute
TRAIN training server (CPU)
GPU hours
—
no GPU used
CPU time
7.3 s
Provider cost
$0.00
nothing purchased

Benchmarks

held-out test set · change vs v0.05
Spotting crashes (−70% within 1h)
0.838
Spotting survivors (still traded after 1h)
0.889
Spotting 2× runs (within 1h)
0.776
Calling the 1h direction (down / flat / up)
47.6%
Spotting creator dumps (within 1h)
0.824
Spotting crashes (within 6h)
insufficient data
Spotting survivors (6h)
insufficient data
Calling the 6h direction
insufficient data
Spotting survivors (24h)
insufficient data
Spotting migrations (within 24h)
insufficient data

Validation metrics (during training)

collapse_1h
AUC 0.962
n=105
collapse_6h
—
n=0
reach_2x_1h
—
n=105
survival_1h
—
n=105
survival_6h
—
n=0
survival_24h
—
n=0
migration_24h
—
n=0
trajectory_1h
acc 98.1%
n=105
trajectory_6h
—
n=0
creator_exit_1h
AUC 0.697
n=44

Validation launches are separate from training launches and from the locked test set. Benchmarks above are the public numbers.

What this version relies on

held-out validation launches
This version was trained before TRAIN started measuring which signals each model relies on.

For each signal, how much the score drops when that signal is scrambled across launches. TRAIN learns these weights itself from the data; nobody hand-codes them.

Training runs

Attempt 1Succeeded
05 Oct 06:00 · 00:00:03 · 679 examples · $0.00

Timeline

  1. 05 Oct 06:01
    TRAIN v0.06 not deployed; v0.01 stays live
    Reason: no shared benchmarks yet; keeping the live model.
  2. 05 Oct 06:01
    Benchmark completed: TRAIN v0.06
    Crash-spotting score: not enough held-out data yet (n=220).
  3. 05 Oct 06:01
    Benchmark started: TRAIN v0.06
  4. 05 Oct 06:01
    Training completed: TRAIN v0.06
    00:00:03 · trained on 679 examples · $0 compute cost
  5. 05 Oct 06:00
    Training started: TRAIN v0.06
    Solana Launch Dataset v5 · decision-tree ensemble · TRAIN training server
  6. 05 Oct 06:00
    TRAIN v0.06 queued
    Adds the realized track record (inside TRAIN's sample, before the snapshot) of the wallets buying the launch — how many are known, profitable, losing or high-frequency.