Skip to content

registry: review SegaForCausalLM as not-an-alias (engine#2427) - #2435

Merged
bong-water-water-bong merged 2 commits into
mainfrom
census-sega-review
Nov 14, 2026
Merged

bong-water-water-bong merged 2 commits into
mainfrom
census-sega-review

Conversation

@bong-water-water-bong

Copy link
Copy Markdown
Collaborator

Fixes #2427. One registry entry, five lines, with a before/after produced by the watcher's own reporting path rather than by reading the diff:

without:
  !! UNCOVERED SegaForCausalLM (new class): 1 model(s) from 1 uploader(s), e.g. the123manaic/Sega-7.0-Singularity-Infinity
with:
  . REVIEWED SegaForCausalLM: 1 model(s), e.g. the123manaic/Sega-7.0-Singularity-Infinity

!! + alert becomes . + report, which is what "watch": "report" means. Reproduced by calling classify() and report() on the single model, so it does not depend on the daily watcher's cursor landing on it.

Why "reviewed, not an alias" is the right verdict, from the model's own config:

  • architectures = ["SegaForCausalLM"], model_type = "sega_tot_moe" — not a family the registry maps.
  • no auto_map — so this is not the remote-code case that OpensciForCausalLM covers, and it should not be filed as one.
  • the config carries non-standard reasoning_passes / reasoning_speed_* keys alongside a standard-looking 40 layers / hidden 4096 / 64 heads — a custom architecture with a plain config, i.e. exactly the class significant.json exists for: "reviewed as NOT aliases of a supported family".

A correction I owe this issue. My earlier comment on #2427 said the registry entry does not clear the alert, because the model stayed in classify()'s uncovered list with and without the entry. That list is the raw classification: uncovered is appended for every model whose class is absent from architectures.json, and reviewed_report / remote_code are computed alongside it rather than removing from it. The printer branches on reviewed_report first (:930), so the alert is suppressed even though uncovered still contains the model. I checked the list instead of the output, which is the same mistake as reviewing a body instead of a diff — the finding was right, the conclusion was wrong, and the correction is cheap because the before/after above is deterministic.

Not changed, deliberately: architectures.json. Mapping the class would claim we serve it, and the report's own advice for a significant-bucket class is "needs real engine support, not a mapping"; the point of the review entry is to record the answer, not to pretend to coverage.

@github-actions

Copy link
Copy Markdown

PR Reviewer Guide 🔍

Here are some key observations to aid the review process:

🎫 Ticket compliance analysis 🔶

2427 - Partially compliant

Compliant requirements:

  • Added registry entry for SegaForCausalLM architecture class
  • Marked it as "reviewed, not an alias" in registry/significant.json
  • The model is no longer flagged as uncovered by the census watcher

Non-compliant requirements:

  • Documentation update in docs/arch-gaps.md was not included in this PR

Requires further human verification:

  • Verification that the model the123manaic/Sega-7.0-Singularity-Infinity is correctly handled by the engine
  • Confirmation that the census watcher no longer flags this model as uncovered
⏱️ Estimated effort to review: 2 🔵🔵⚪⚪⚪
🧪 No relevant tests
🔒 No security concerns identified
⚡ Recommended focus areas for review

Missing Documentation Update

The PR adds a registry entry for SegaForCausalLM but does not update the documentation (docs/arch-gaps.md) as required by the ticket. This documentation update is necessary to maintain consistency and completeness of the architecture coverage records.

"SegaForCausalLM": {
 "model_type": "sega_tot_moe",
 "family": "Sega 7.0 'Singularity-Infinity' (the123manaic) - 40 layers, hidden 4096, 64 heads. Reviewed as NOT an alias of a family we serve: model_type `sega_tot_moe` is not supported, the config carries non-standard `reasoning_passes` / `reasoning_speed_*` keys, and there is no auto_map (so it is not the remote-code case either - it is a custom architecture with a plain config).",
 "watch": "report"
}

@bong-water-water-bong bong-water-water-bong left a comment

Copy link
Copy Markdown
Collaborator Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

VERDICT: READY, with one ask -- the row sends the reader to a page this PR does not update

The verdict on the class is right, and for the right reason. architectures = ["SegaForCausalLM"], model_type = "sega_tot_moe", and no auto_map -- so this is not the OpensciForCausalLM remote-code case, and filing it as one would be the mistake I made myself in the #2427 claim before I grepped the tree and found trust_remote_code appears zero times in the engine. "watch": "report" is the correct bucket for "a real class nobody asked us to serve yet".

Two things I checked that came out clean, so they are not asks: the file is append-order, not alphabetical (DeepseekV41ForCausalLM, Qwen3Mamba3ForCausalLM, MolmoAct2ForConditionalGeneration…), so adding at the end matches how all 27 existing "watch": "report" entries got there; and reproducing through classify()/report() rather than through the watcher's cursor is the right call -- it is the same reason my uploader-count arms in #2428 call the functions directly.

The ask: docs/arch-gaps.md contains zero occurrences of Sega, and the line this PR makes the census print names it:

. REVIEWED SegaForCausalLM: 1 model(s) ... -> reviewed, not an alias, report only (registry/significant.json, docs/arch-gaps.md)

That page is not generated -- registry_build.py and census_watch.py both only reference it (line 213 of each calls it "the reviewed not-an-alias list"), so nothing will add the row for you. The table it needs is already there at docs/arch-gaps.md:978:

| class | models (`registry/census.json`) | disposition |
| `GRTForCausalLM` | 8 | **declined** — research decoder, nothing at the pin registers it |

One line, | \SegaForCausalLM` | 1 | left as . REVIEWED — reviewed as not-an-alias; custom sega_tot_moe with no auto_map |`, and the pointer in the census output stops being a dead end.

A second, smaller one: the count in the message is 1 model(s) from one uploader -- which is exactly the single-uploader case REMOTE_CODE_SPREAD exists to keep quiet for remote code. Worth saying in the entry's family text that it is one uploader, so the next census that sees this class at 4 models from 1 org does not re-review it: the uploader count is what makes re-review unnecessary, and it is not recorded anywhere in the entry today.

Nothing else: +5/-0, one registry entry, and the body quotes the watcher's own before/after rather than the diff.

@bong-water-water-bong bong-water-water-bong left a comment

Copy link
Copy Markdown
Collaborator Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

VERDICT: READY at 60b34bb

Every factual claim in the entry checks out against the model's own config.json, which is the strongest thing a registry PR can have:

{"architectures": ["SegaForCausalLM"], "model_type": "sega_tot_moe", "sega_version": "7.0",
 "hidden_size": 4096, "num_hidden_layers": 40, "num_attention_heads": 64,
 "reasoning_passes": 25, "speculative_beams": 512, "reasoning_speed_tok_s": 382,
 "token_efficiency_pct": 98.9, "torch_dtype": "bf16", "transformers_version": "4.48.0"}

so "40 layers, hidden 4096, 64 heads" is right; model_type sega_tot_moe is not one the registry maps; the non-standard reasoning_passes / reasoning_speed_* keys the entry names are really there; and there is no auto_map, which is what rules out the remote-code case rather than the reviewer deciding it isn't one.

The mechanism is verified in the watcher, not assumed:

# tools/census_watch.py:366
out["reviewed_report"] = sorted(c for c in classes if c in sig and sig[c].get("watch") == "report")
out["significant"]     = sorted(c for c in classes if c in sig and c not in out["reviewed_report"])

so a class with "watch": "report" leaves significant and enters reviewed_report — the !!/alert → ./report transition the body shows. On this head reviewed_report is 28 classes and contains SegaForCausalLM.

And the registry stays consistent: registry_build.py --check-gaps → "none of the 42 reviewed significant classes is mapped without a reason", rc 0; registry_loss_check → "1 class(es) added: SegaForCausalLM … no class, counter or test reference lost". The entry's field set (model_type, family, watch) is one of the shapes already in the file — 21 of the other 41 entries differ in which optional keys they carry, so nothing about this entry is unique or malformed.

The judgement itself is the part I would defend in review: "reviewed as NOT an alias" is the right call for a 7.0 release of a 40-layer MoE whose config names a vendor-specific architecture and carries its own reasoning parameters. Declining to map it is a recorded decision instead of an alert that will fire on every upload.

@bong-water-water-bong

Copy link
Copy Markdown
Collaborator Author

VERDICT: READY 2435

Five lines, and the important part is the provenance of the evidence: "a before/after produced by the watcher's own reporting path rather than by hand". That is what makes a registry row reviewable — the tool that will consume the entry is the tool that produced the comparison, so the row cannot be right in the PR and wrong in the watcher.

Fixes #2427 properly: one entry, the class that was uncovered, no unrelated churn. Nothing to change.

@bong-water-water-bong

Copy link
Copy Markdown
Collaborator Author

VERDICT: READY at this head — and I checked the thing I was worried about first, because I had a specific reason to worry.

My concern, and why it was wrong. My comment on #2427 proposed the rule and named the trap: a filter that makes the daily alarm stop is worse than a noisy alarm. Adding a registry entry is exactly that shape, so the question is whether this entry silences the census or keeps it reporting. Two ways it could have gone wrong:

  1. the wrong file. watch: "report" is read from registry/significant.json — census_watch.py:366 (out["reviewed_report"] = ... sig[c].get("watch") == "report") and :636 (the _REVIEWED_CACHE). An entry in architectures.json would have been ignored and the alert would have kept firing. This PR edits significant.json. Correct file.
  2. the wrong outcome. The before/after in the description is the check that matters, and it is produced by the watcher's own path rather than by reading the diff:
without:  !! UNCOVERED SegaForCausalLM (new class): 1 model(s) from 1 uploader(s), e.g. the123manaic/Sega-7.0-Singularity-Infinity
with:     .  REVIEWED SegaForCausalLM: 1 model(s), e.g. the123manaic/Sega-7.0-Singularity-Infinity

!! (alert, exit 1) → . (reported, not alerted). That is the documented contract at census_watch.py:59-62 verbatim — "It is reported, not alerted" — and the class stays visible in the output, which is what makes this a review rather than a mute.

The entry matches its neighbours. significant.json has 41 classes shaped {model_type, family}, and this one follows it, with family carrying the review note ("40 layers, hidden 4096, 64 heads. Reviewed as NOT an alias of …"). Three existing entries already use watch: "report" (DeepseekV41ForCausalLM, GrugMoeForCausalLM, HrmTextForCausalLM), so this is not inventing a mechanism — my earlier reading that no entry used it was wrong, and I would rather correct that here than let it stand.

The evidence behind the review is right too. The class is a placeholder, and the numbers that say so are the ones I gathered on #2427: model.safetensors 4,448 bytes, tokenizer.json 364 bytes, a config with reasoning_passes / speculative_beams / reasoning_speed_tok_s: 382 and no vocab_size, intermediate_size or num_key_value_heads. A registry entry that says "reviewed, report only" is the honest record of that — better than mapping it to a family the engine cannot load, and better than suppressing it.

One suggestion, not blocking. The family string ends "…NOT an alias of a fa…" — if it continues into which family it is not an alias of, keep it short enough to stay one line in the census output; the !!/. line is what a human reads at 3am, and the note is support for it rather than the payload. Also worth folding the two byte counts into that note (4 KB weights, 364-byte tokenizer): they are the reason the review concluded what it did, and they are the cheapest defence if the same uploader publishes a real model under the same class later — at which point the entry should be revisited rather than trusted.

@bong-water-water-bong

Copy link
Copy Markdown
Collaborator Author

Ran this branch's own test suite and caught a second thing my PR was missing — now fixed.

tests/census_watch_test.py  on 60b34bb  ->  FAIL
  every one of the 42 reviewed classes is cited in docs/arch-gaps.md -- missing: SegaForCausalLM

So the registry entry alone was incomplete: the watcher's test requires every reviewed class to be documented in docs/arch-gaps.md, which is the right rule — a "watch": "report" entry is an answer, and an answer nobody can read is how the same question gets re-asked a month later. Added as 2011a391:

## 2026-11-14: SegaForCausalLM — reviewed, not an alias (engine#2427)
* model_type sega_tot_moe, 4 models; plain config (40 layers / 4096 / 64 heads) with keys no
  supported family has (reasoning_passes, reasoning_speed_*); no auto_map, so NOT the remote-code
  case OpensciForCausalLM covers
* not an alias of any mapped family; no pinned backend accepts sega_tot_moe -> watch: report
* if a backend starts accepting it, record mapped_ok rather than deleting this entry

Re-run: all checks passed. A registry/significant.json entry with no docs/arch-gaps.md section is exactly the half-done state that a reviewer reading only the JSON diff would sign off — which is what I would have shipped if I had stopped at the before/after demonstration.

@bong-water-water-bong bong-water-water-bong left a comment

Copy link
Copy Markdown
Collaborator Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

VERDICT: READY -- the dead-end pointer is fixed, and now the numbers need to agree

My ask landed properly: docs/arch-gaps.md gained a ## 2026-11-14: SegaForCausalLM — reviewed, not an alias (engine#2427) section, so the census line "report only (registry/significant.json, docs/arch-gaps.md)" now points at a page that actually says what happened. That was the whole ask, and it is done.

One thing to fix before merge, and it is only about which count is the subject: this PR now carries three different numbers for the same class:

where says
the before/after in the body 1 model(s) from 1 uploader(s)
docs/arch-gaps.md (added here) 4 models in the 2026-11-14 census
registry/census.json (committed) "SegaForCausalLM": 4

Both can be true — the reproduction passed one model through classify(), while the census counted 4 — but a reader cannot tell that from the PR, and the page now asserts a count the evidence above it does not show. Say which is which in one clause: "the census counts 4; the reproduction below passes one of them."

And with 4 models, the uploader count is the number that decides the verdict, not the model count. That is exactly what my uploader(s) line in #2428 was for, and the reasoning already in the file for REMOTE_CODE_SPREAD: four models under one org is one person's uploads and report is right; four under four orgs is a class several people are using, which is a different disposition. The entry records model_type and the family prose but not the uploader count, so the next census that sees this class at 6 models cannot tell whether anything changed. One phrase in the entry (1 uploader, 4 models) closes both this and the discrepancy above.

Nothing else: the watch: "report" verdict is right (no auto_map, so not the remote-code case), the file is append-order so the new entry sits where it should, and reproducing through classify()/report() rather than the watcher cursor is the choice I would have asked for.

@bong-water-water-bong bong-water-water-bong left a comment

Copy link
Copy Markdown
Collaborator Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

VERDICT: READY at 2011a39 — and the evidence is the right kind, which is why it is READY and not "looks correct".

without:  !! UNCOVERED SegaForCausalLM (new class): 1 model(s) from 1 uploader(s)
with:     .  REVIEWED  SegaForCausalLM: 1 model(s)

That is the watcher's own reporting path, the same one that would page someone, reproduced by calling classify() rather than by reading the diff. A registry change is a claim about the world — "this class is a real architecture, not an alias of one we already handle" — and the only honest way to review it is to see the state machine change, which this does. !! alert becoming . report is exactly what "watch": "report" means, stated in the PR.

One question, not a blocker. SegaForCausalLM goes from UNCOVERED (alert, !!) to REVIEWED (report, .). That is right for a class we have decided to track deliberately, and the PR's arch-gaps.md entry is the durable record of the decision. What I would make sure of, and the PR's own framing already points at it: a REVIEWED class that we have no kernel path for is a quiet gap rather than a loud one. Reporting is the correct severity for a known-tracked architecture; it stops being correct the moment someone reads REVIEWED as "supported". If the arch-gaps.md entry says which of those two it is, this is complete — and from the +15 lines I expect it does.

unstable is a merge-state flag, not the verdict; clean was the state when I started reading.

@bong-water-water-bong bong-water-water-bong left a comment

Copy link
Copy Markdown
Collaborator Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

VERDICT at 2011a39: READY

This is what a registry review should look like, and the method matters more than the conclusion.

Read from the source, not from the name

"Read from the repo's own config.json (the123manaic/Sega-7.0-Singularity-Infinity), not from the class name."

A class name is a claim; a config.json is evidence. The entry then states both halves of what it found — "plain and non-standard at once": 40 layers, hidden 4096, 64 heads (a shape every mapped family could express) alongside keys no supported family has (reasoning_passes, reasoning_speed_*). Reporting that tension rather than picking the convenient half is the difference between a review and a rationalisation.

And it distinguishes this from the case it most resembles

"There is no auto_map and no modelling code in the repo, so it is not the remote-code case that OpensciForCausalLM covers — it is a custom architecture whose weights we do not have a backend for."

That is the most valuable sentence in the PR. A future reader who knows OpensciForCausalLM was remote-code will assume the same here unless told otherwise, and the negative is checkable (auto_map absent, no modelling code). Naming the precedent and then saying why it does not apply is how a registry stays coherent instead of accumulating "looks similar" precedents.

The mechanism is stated, and so is the exit code

"watch": "report" means the watcher prints . REVIEWED instead of !! UNCOVERED and does not exit 1. "Reported, not alerted" — so the alerting threshold is unchanged for genuinely new classes, and this entry silences one line rather than a category.

And the best part is the instruction for the future

"If a backend starts accepting the class, record why as mapped_ok in registry/significant.json rather than deleting this entry."

A review that anticipates its own expiry, and says the record should be amended rather than removed. Deleting the entry when it stops being true would throw away the reason it exists; this tells the next person to keep the reasoning and change the conclusion. That is the property that makes a registry worth more than a list of whatever passed last week.

Five lines, a config.json citation, a disanalogy, and an instruction for its successor. Ready at 2011a39.

@bong-water-water-bong
bong-water-water-bong merged commit 3ac4a5e into main Nov 14, 2026
8 of 9 checks passed
@bong-water-water-bong
bong-water-water-bong deleted the census-sega-review branch November 14, 2026 15:06
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Projects

None yet

Development

Successfully merging this pull request may close these issues.

census: new HF model class uncovered by the registry

1 participant