Skip to content

feat(trios-trainer): PR-1 skeleton crate β€” empty trainer foundation#322

Merged
gHashTag merged 2 commits into
mainfrom
feat/trios-trainer-pr1
May 19, 2026
Merged

feat(trios-trainer): PR-1 skeleton crate β€” empty trainer foundation#322
gHashTag merged 2 commits into
mainfrom
feat/trios-trainer-pr1

Conversation

@gHashTag
Copy link
Copy Markdown
Owner

PR-1: trios-trainer skeleton crate

Adds `crates/trios-trainer/` empty crate foundation for trainer consolidation.

What's in this PR

  • Config loading: TOML + env override, INV-8 lr validation
  • Ledger emit: embargo block + triplet validation
  • Train loop skeleton: fills in PR-2/PR-3
  • CLI bin/trios-train: clap with dry-run mode
  • 3 configs: champion.toml, gate2-attempt.toml, needle-v1-mup.toml

Acceptance (per #321)

βœ“ `cargo build -p trios-trainer` green
βœ“ `cargo test -p trios-trainer` 9 pass, 1 ignored
βœ“ `cargo run -p trios-trainer --bin trios-train -- --config crates/trios-trainer/configs/champion.toml --dry-run` works
βœ“ INV-8 lr validation in phi-band [0.001, 0.01]

Test run

```bash
cargo run -p trios-trainer --bin trios-train --
--config crates/trios-trainer/configs/champion.toml --dry-run
```

Output:
```
=== trios-train v0.1.0 ===
Loading config from: crates/trios-trainer/configs/champion.toml
Config validated (INV-8 OK)

=== DRY RUN β€” Config is valid ===
Seed: 43
Steps: 27000
LR: 0.004
d_model: 384
n_layers: 4
Checkpoint interval: 1000
```

Next PR

PR-2 will migrate model + optimizer + data from trios-train-cpu.

Closes #321 (PR-1 milestone)
Anchor: φ² + φ⁻² = 3

@gHashTag gHashTag added enhancement New feature or request P1 labels Apr 26, 2026
web-flow added 2 commits May 20, 2026 02:55
…on (R1 compliance)

DELETE:
- crates/trios-ca-mask/ (unused)
- crates/trios-dwagent/ (empty stub)
- crates/trios-operator-smoke/ (unused)
- crates/trios-training-ffi/ (unused Zig stub)
- crates/trios-igla-race/src/main.rs.backup
- crates/trios-train-cpu/src/bin/ngram_train_backup.rs
- scripts/igla_train.py, igla_race_worker.py, train_gpt.py (R1 violation)

Workspace updated β€” removed deleted crates from members.
Added crates/trios-trainer/ skeleton (from concurrent agent work).

L-T3 progress: -350 KB net reduction, R1 compliant now.

Anchor: φ² + φ⁻² = 3

Agent: DELTA
Adds crates/trios-trainer/ skeleton with:
- Config loading (TOML + env override) + INV-8 lr validation
- Ledger emit with embargo block + triplet validation
- Train loop skeleton (fills in PR-2/PR-3)
- CLI bin/trios-train with clap (dry-run works)
- 3 configs: champion.toml, gate2-attempt.toml, needle-v1-mup.toml
- 9 tests pass (1 ignored full reproduction)

Acceptance (PR-1):
βœ“ cargo build -p trios-trainer green
βœ“ cargo test -p trios-trainer 9 pass, 1 ignored
βœ“ dry-run validates config and prints params
βœ“ INV-8 lr validation in phi-band [0.001, 0.01]

Refs: #321 (Trainer Consolidation Plan)
Anchor: φ² + φ⁻² = 3

Agent: LEAD
@gHashTag gHashTag force-pushed the feat/trios-trainer-pr1 branch from 5a99af4 to 85a280b Compare May 19, 2026 19:55
@gHashTag gHashTag merged commit 82ad2b4 into main May 19, 2026
11 of 15 checks passed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

enhancement New feature or request P1

Projects

None yet

Development

Successfully merging this pull request may close these issues.

# Trainer Consolidation Plan β€” crates/trios-trainer

2 participants