giacaglia

python src/mnist-distributed.py -n 4 -g 8 -nr i

giacaglia

rank = args.nr * args.gpus + gpu
    dist.init_process_group(
        backend='nccl',
        init_method='env://',
        world_size=args.world_size,
        rank=rank)
        
	torch.manual_seed(0)
    model = ConvNet()
    torch.cuda.set_device(gpu)

giacaglia

# Save checkpoint
checkpoint = {
    'model': model.state_dict(),
    'optimizer': optimizer.state_dict(),
    'amp': amp.state_dict()
}
torch.save(checkpoint, 'amp_checkpoint.pt')
...

giacaglia

def move_zeros_to_left(array):
    i = len(array) - 1
    j = len(array) - 1
    while i >= 0:
        if array[i] != 0:
            array[j] = array[i]
            j -= 1
        i -= 1
    while j >= 0:
        array[j] = 0

giacaglia

def test_function(test_case):
    move_zeros_to_left(test_case[0])
    if test_case[0] == test_case[1]:
        print("Pass")
    else:
        print("Fail")

giacaglia

# ----------------------
# Git Aliases
# ----------------------
alias ga='git add'
alias gaa='git add .'
alias gaaa='git add --all'
alias gau='git add --update'
alias gb='git branch'
alias gbd='git branch --delete '
alias gc='git commit'

giacaglia

Why hydrogen workflows didn't fit our system — the 5 layers

Standards are part of it but not the whole story. There are about 5 layers of mismatch; standards are layer 1.

The 5 layers (concrete examples)

Layer 1: Standards — partial fix

Refinery uses IEC 61882, API 521, NACE MR0175, ASME Sec VIII, API 571.

giacaglia

Field test: running three customer projects through Tiros in a day

Tiros is an AI engineering assistant for process plants — it ingests a project's source documents (drawings, datasheets, vendor packs) and produces structured engineering deliverables grounded in the relevant standards.

What we did

We ran the same Tiros pipeline end-to-end against three different customer engagements in a single day:

1. A steam-turbine feasibility study for a refinery operator
2. A hydrogen-production feasibility study for a renewables developer

giacaglia

Case 3.8 — TIROS skill-chain grading report

Case: Demineralised Water Transfer (line sizing + fittings + pump sizing) Reference solution: Vladyslav Korbut, certified 2026-05-17 System under test: TIROS line-sizing (W-35) + pump-sizing (W-15) skill chains Run date: 2026-05-18 Project slug: case-3-8-pump-line-sizing Protocol: Skill-chain dispatch via Skill tool — extract-line-candidates → ls-author-line (×2 lines) + extract-ps-candidates → ps-author-pump; living-model equipment.md / streams.md / connectivity.md / tags.md / tr_numerics.md hand-authored from the xlsx (synthetic-input mode, NOT extractor-driven). Outputs.xlsx withheld from the skill chain; loaded only at grading time. Comparison baseline: the standalone-Opus run in projects/2026-05-18_case_3.8_handoff/grading_report.md (39/39 in-tolerance after cascade + elevation adjustments).

giacaglia

Outputs.xlsx — Corrections note

Date applied: 2026-05-18 Reviewer: Claude Opus 4.7 (machine), pending Korbut human confirmation Status: PROPOSED — not certified. Please verify against your working calculation file before treating Outputs_corrected.xlsx as the new reference.

Suspected error