Home / NandhaKishorM / laya
AI Slop Code RIsk Monitor

NandhaKishorM/laya

Non-autoregressive System 1 decision engine. Typed choice, score and yes/no decisions over any text in a single forward pass, in 100+ languages, with a router that picks the right checkpoint per request.

8.8 Moderate pattern signal View on GitHub

Maintenance and Governance Review

This report summarizes scanner evidence for NandhaKishorM/laya, a Python project with 32,341 GitHub stars. SynthScan examined 376,213 lines of code across 1004 source files, recording 2104 pattern matches distributed across 19 syntactic categories. The legacy adjusted pattern signal is 8.8 (Moderate pattern signal); it is not an authorship estimate.

Use the findings to ask whether maintainers can explain intent, verify dependencies, simplify unnecessary scaffolding, and focus review effort. Risk Payload v1 adds ownership, supply-chain, over-engineering, and review-effort vectors when available. These are prompts for human verification, not probabilities, security findings, or measured maintenance costs.

8.8
Adjusted Signal
8.8
Raw Pattern Score
100%
Time Factor
2026-10-11
Last Push
32.3K
Stars
Python
Language
376.2K
Lines of Code
1.0K
Files
2.1K
Pattern Hits
2026-10-12
Scan Date
0.29
HC Hit Rate

Risk Review · v3

Use each vector to guide manual verification of the underlying findings.

Cognitive Ownership Risk 15.7/100 · LOW

Can maintainers explain intent, assumptions, and trade-offs? 1342 finding(s).

Hallucinated Dependency & Supply Chain Risk 14.3/100 · LOW

Are imports declared and package/API names verified? 349 finding(s).

Agent Over-Engineering Ratio 20.4/100 · LOW

Are wrappers, guards, and repeated structures necessary? 1370 finding(s).

PR Review Velocity Impact 41.2/100 · MEDIUM

Which changes need focused human review? 2104 finding(s).

What These Metrics Mean

Adjusted Signal
Legacy pattern score normalized per 1,000 scanned lines and multiplied by the repository-level time factor. It is not an estimate of AI authorship.
Raw Pattern Score
Sum of severity- and context-weighted pattern findings before applying the time factor.
Time Factor
A coarse multiplier based on repository last-push metadata. It does not identify when individual code was written.
Pattern Hits
Count of configured pattern findings. Review each location and surrounding code before acting.
HC Hit Rate
High- and critical-severity findings per scanned file. Severity is a configured detector weight, not proof of a vulnerability or authorship.
Lines / Files
Scan coverage denominators used by the legacy normalized score.

Pattern-Signal History

Run additional scans to build a history of the legacy pattern signal. Score changes are not a measure of AI adoption or code quality.

No multi-scan history yet — run the scanner again to build trend data.

Severity Breakdown

Groups findings by configured severity weight. Severity is not a confidence estimate, security rating, or proof that a package is missing. Verify imports against the project's manifests and inspect each finding in its local design context.

CRITICAL 16HIGH 277MEDIUM 183LOW 1628

Directory Score Breakdown

This horizontal bar chart shows where configured pattern scores accumulate across directories. Treat the bars as locations to inspect, not as evidence of authorship or code quality. Use the file-level findings to decide whether a directory needs a focused ownership, dependency, complexity, or review check.

Pattern Findings

The scanner identified 2104 distinct pattern matches across 19 syntactic categories. Each entry below represents a discrete location in the source code where the engine recorded a detector match to inspect in context. Expand any category row to inspect the individual file paths, line numbers, code snippets, and the lexical context (CODE, COMMENT, or STRING) in which each match was detected.

Reading the findings table: The Severity column indicates the configured finding weight (CRITICAL / HIGH / MEDIUM / LOW). The Context column identifies whether the match occurred inside executable code, an inline comment, or a string literal — legacy comment-context matches receive a ×1.5 score weight. The ⚡ bolt icon marks clustered matches: three or more patterns within a 10-line window, each receiving an additional ×1.5 density multiplier under the legacy scoring model.

Cross-Language Confusion225 hits · 898 pts
SeverityFileLineSnippetContext
HIGHresearch/evals/test_act_head_eval.py76# One class absent: ROC-AUC is undefined, and must not be faked as 0.5 or 0.0.COMMENT
HIGHlaya/serve.py841 # `{"<i>": null}`, which a Jev client's schema refuses to parse (#302). Reject it as aCOMMENT
HIGHlaya/agent.py1031 "question %r: choice label %d is null; a label is rendered as option text "CODE
HIGHlaya/agent.py1068 raise ValueError("question %r: score level %d is null; give every level a description, "CODE
HIGHlaya/mcp/server.py115 "flagged 'low_confidence' (and, on laya_decide, its value comes back null), so a caller that "CODE
HIGHtests/test_packaging.py693 check_true("nix/%s is opt-in (nullOr, default null)" % opt,CODE
HIGHtests/test_packaging.py694 _t.count("nullOr") == 1 and re.search(r"^\s*default = null;", _t, re.M) is not None,CODE
HIGHtests/test_packaging.py799_PER_COMMIT = "github.event_name == 'pull_request' && github.ref || github.sha"CODE
HIGH⚡tests/test_serve.py2572 """`{"hooks": null}` means "no per-call hooks", which is what core reads as inherit.STRING
HIGHtests/test_serve.py1250 """A null score level would come back as `legend: {"<i>": null}`, which Jev clients refuse toSTRING
HIGHtests/test_serve.py2743 """``{"lang_guess": null}`` means "no hint" -- same shape the single endpoint honours.STRING
HIGH⚡examples/server.py952 var r = document.documentElement, t = null;CODE
HIGH⚡examples/server.py954 var pick = function (k) { return k && Object.prototype.hasOwnProperty.call(names, k) ? names[k] : null; };CODE
HIGH⚡examples/server.py954 var pick = function (k) { return k && Object.prototype.hasOwnProperty.call(names, k) ? names[k] : null; };CODE
HIGH⚡examples/server.py958 if (q || t) r.dataset.theme = q || t; /* runs in <head>, before first paint */CODE
HIGH⚡examples/server.py993 while (box && box !== document.body && getComputedStyle(box).overflow === "visible") box = box.parentElement;CODE
HIGH⚡examples/server.py994 const r = w.getBoundingClientRect(), g = box && box !== document.body ? box.getBoundingClientRect() : null;CODE
HIGH⚡examples/server.py994 const r = w.getBoundingClientRect(), g = box && box !== document.body ? box.getBoundingClientRect() : null;CODE
HIGH⚡examples/server.py1710const $ = (sel, root) => (root || document).querySelector(sel);CODE
HIGH⚡examples/server.py1716 if (v == null || v === false) continue;CODE
HIGH⚡examples/server.py1723 for (const c of kids.flat(Infinity)) if (c != null && c !== false) el.append(c);CODE
HIGH⚡examples/server.py1739 try { const v = localStorage.getItem(k); return v == null ? null : JSON.parse(v); } catch (e) { return null; }CODE
HIGH⚡examples/server.py1744const isMap = (x) => !!x && typeof x === "object" && !Array.isArray(x);CODE
HIGH⚡examples/server.py1751const plainEnter = (e) => e.key === "Enter" && !e.isComposing && e.keyCode !== 229 && !e.shiftKey && !e.ctrlKeyCODE
HIGH⚡examples/server.py1856 if (!node || node.t !== "obj") return null;CODE
HIGH⚡examples/server.py1856 if (!node || node.t !== "obj") return null;CODE
HIGH⚡examples/server.py1858 return null;CODE
HIGH⚡examples/server.py1862 ind = ind || 0;CODE
HIGH⚡examples/server.py1870const pretty = (v) => JSON.stringify(v, null, 2);CODE
HIGH⚡examples/server.py1894 const errAt = marks && marks.errAt != null ? marks.errAt : -1;CODE
HIGH⚡examples/server.py1895 const warn = (marks && marks.warn) || new Set();CODE
HIGH⚡examples/server.py1899 const end = off + line.length, errHere = errAt >= off && errAt <= end;CODE
HIGH⚡examples/server.py1905 if (errHere && errAt >= at && errAt < at + t.length) {CODE
HIGH⚡examples/server.py1912 if (errHere && errAt === end) div.append(span("", "eol"));CODE
HIGH⚡examples/server.py1990 const pane = sc && getComputedStyle(sc).overflowY !== "visible" ? sc : null;CODE
HIGH⚡examples/server.py1990 const pane = sc && getComputedStyle(sc).overflowY !== "visible" ? sc : null;CODE
HIGH⚡examples/server.py1998 if (dir > 0) while (li < rows.length - 1 && off(li + 1) <= off(li)) li++;CODE
HIGH⚡examples/server.py1999 else while (li > 0 && off(li - 1) <= off(li)) li--;CODE
HIGH⚡examples/server.py2005 (pane || window).scrollBy(0, dy);CODE
HIGH⚡examples/server.py2037 else if (e.altKey && e.shiftKey && e.code === "KeyF") { e.preventDefault(); ed.format(); }CODE
HIGH⚡examples/server.py2043 if (!out && s === e) { insertText(ta, " "); return; }CODE
HIGH⚡examples/server.py2045 let le = v.indexOf("\n", e > s && v[e - 1] === "\n" ? e - 1 : e);CODE
HIGH⚡examples/server.py2077const MOD = /Mac|iPhone|iPad/.test(navigator.platform || navigator.userAgent) ? "\u2318" : "Ctrl";CODE
HIGH⚡examples/server.py2080const SOURCES = Object.create(null); /* no inherited keys, so ?preset=constructor is just unknown */CODE
HIGH⚡examples/server.py2086 stateMode: "fields", qMode: "form", fields: [], qs: [], base: null, last: null, running: false,CODE
HIGH⚡examples/server.py2087 problems: [], body: null, qValue: null, view: "answers", notes: {state: null, questions: null},CODE
HIGH⚡examples/server.py2146 if (!t || t.v.t !== "str" || !FORM_TYPES.includes(t.v.v))CODE
HIGH⚡examples/server.py2148 if (ins && ins.v.t !== "str") return "The instructions of " + name + " are not text.";CODE
HIGH⚡examples/server.py2149 if (!c || (c.v.t === "lit" && c.v.v === null)) continue;CODE
HIGH⚡examples/server.py2150 const textOrNull = (x) => x.t === "str" || (x.t === "lit" && x.v === null);CODE
HIGH⚡examples/server.py2157 if (t.v.v === "noul" && !(tf.length && tf.every((k) => k === "true" || k === "false") &&CODE
HIGH⚡examples/server.py2158 new Set(tf).size === tf.length && c.v.entries.every((o) => textOrNull(o.v))))CODE
HIGH⚡examples/server.py2166 opts: [], levels: [], tf: null, extra: []};CODE
HIGH⚡examples/server.py2168 obj.entries.map((o) => ({label: o.k, desc: o.v.v == null ? "" : o.v.v, wasNull: o.v.v === null}));CODE
HIGH⚡examples/server.py2169 if (c && q.type === "choice" && c.v.t === "obj") q.opts = pairs(c.v);CODE
HIGH⚡examples/server.py2170 if (c && q.type === "choice" && c.v.t === "arr") { /* a list of labels stays a list until one gets a description */CODE
HIGH⚡examples/server.py2174 if (c && q.type === "noul" && c.v.t === "obj") q.tf = pairs(c.v); /* what true and false mean */CODE
HIGH⚡examples/server.py2175 if (c && q.type === "score" && c.v.t === "arr") q.levels = c.v.items.map((x) => x.v);CODE
HIGH⚡examples/server.py2177 for (const e of v.entries) if (!seen.has(e.k)) { seen.add(e.k); q.extra.push([e.k, toValue(lastEntry(v, e.k).v)]); }CODE
HIGH⚡examples/server.py2185 const asList = q.optList && opts.every((x) => x.desc === "");CODE
165 more matches not shown…
Hyper-Verbose Identifiers922 hits · 812 pts
SeverityFileLineSnippetContext
LOWresearch/benchmarks/feishu_zh/tests/test_audit.py13 def test_archived_source_bytes(self):CODE
LOWresearch/benchmarks/feishu_zh/tests/test_audit.py35 def test_reject_corrupted_records(self):CODE
LOWresearch/benchmarks/feishu_zh/tests/test_audit.py49 def test_failed_request_stays_in_denominator(self):CODE
LOWresearch/benchmarks/feishu_zh/tests/test_audit.py60 def test_balanced_frozen_cases(self):CODE
LOWresearch/benchmarks/feishu_zh/tests/test_audit.py65 def test_perfect_and_failed_classification(self):CODE
LOW⚡…benchmarks/es_phone_turns/tests/test_es_phone_turns.py29 def test_answering_keep_talking_by_mistake_is_not_a_wrong_action(self):CODE
LOW⚡…benchmarks/es_phone_turns/tests/test_es_phone_turns.py34 def test_acting_when_it_should_not_is_a_wrong_action(self):CODE
LOW⚡…benchmarks/es_phone_turns/tests/test_es_phone_turns.py39 def test_the_gate_only_lets_actions_through(self):CODE
LOW⚡…benchmarks/es_phone_turns/tests/test_es_phone_turns.py49 def test_borderline_cases_stay_out_of_the_headline(self):CODE
LOW⚡…benchmarks/es_phone_turns/tests/test_es_phone_turns.py59 def test_every_rung_asks_the_same_question_with_the_same_options(self):CODE
LOW⚡…benchmarks/es_phone_turns/tests/test_es_phone_turns.py66 def test_every_case_has_a_label_of_its_menu(self):CODE
LOW⚡…benchmarks/es_phone_turns/tests/test_es_phone_turns.py75 def test_no_training_destination_is_a_test_destination(self):CODE
LOW⚡…benchmarks/es_phone_turns/tests/test_es_phone_turns.py83 def test_a_test_sentence_with_one_word_changed_is_refused(self):CODE
LOW⚡…benchmarks/es_phone_turns/tests/test_es_phone_turns.py91 def test_a_case_always_offers_its_own_answer(self):CODE
LOW…benchmarks/es_phone_turns/tests/test_es_phone_turns.py107 def test_the_committed_training_set_holds_no_test_sentence(self):CODE
LOW⚡…benchmarks/es_phone_turns/tests/test_es_phone_turns.py144 def test_the_untouched_archive_passes(self):CODE
LOW⚡…benchmarks/es_phone_turns/tests/test_es_phone_turns.py147 def test_a_better_accuracy_than_the_records_give_is_caught(self):CODE
LOW⚡…benchmarks/es_phone_turns/tests/test_es_phone_turns.py152 def test_a_flipped_decision_is_caught(self):CODE
LOW⚡…benchmarks/es_phone_turns/tests/test_es_phone_turns.py158 def test_a_rewritten_gold_is_caught(self):CODE
LOW⚡…benchmarks/es_phone_turns/tests/test_es_phone_turns.py166 def test_a_dropped_case_is_caught(self):CODE
LOW⚡…benchmarks/es_phone_turns/tests/test_es_phone_turns.py171 def test_a_hidden_wrong_action_is_caught(self):CODE
LOW⚡…h/benchmarks/zh_short_commands/tests/test_benchmark.py46 def test_cases_parse_and_are_unique(self):CODE
LOW⚡…h/benchmarks/zh_short_commands/tests/test_benchmark.py51 def test_every_case_has_a_known_gold_label(self):CODE
LOW⚡…h/benchmarks/zh_short_commands/tests/test_benchmark.py57 def test_every_label_is_covered(self):CODE
LOW⚡…h/benchmarks/zh_short_commands/tests/test_benchmark.py63 def test_choice_options_are_the_label_policy(self):CODE
LOW⚡…h/benchmarks/zh_short_commands/tests/test_benchmark.py66 def test_four_noul_dimensions(self):CODE
LOW⚡…h/benchmarks/zh_short_commands/tests/test_benchmark.py71 def test_noul_criteria_use_the_two_documented_keys(self):CODE
LOW⚡…h/benchmarks/zh_short_commands/tests/test_benchmark.py140 def test_json_state_is_the_only_wrapped_state(self):CODE
LOW⚡…h/benchmarks/zh_short_commands/tests/test_benchmark.py149 def test_noul_rungs_ask_the_four_dimensions_in_order(self):CODE
LOW⚡…h/benchmarks/zh_short_commands/tests/test_benchmark.py156 def test_choice_rungs_ask_one_six_option_question(self):CODE
LOW⚡…h/benchmarks/zh_short_commands/tests/test_benchmark.py164 def test_unknown_config_is_rejected(self):CODE
LOW⚡…h/benchmarks/zh_short_commands/tests/test_benchmark.py236 def test_accuracy_re_derives_from_the_cases(self):CODE
LOW⚡…h/benchmarks/zh_short_commands/tests/test_benchmark.py244 def test_noul_accuracy_re_derives_per_dimension(self):CODE
LOW⚡…h/benchmarks/zh_short_commands/tests/test_benchmark.py253 def test_hashes_of_the_frozen_inputs_are_recorded(self):CODE
LOW⚡…h/benchmarks/zh_short_commands/tests/test_benchmark.py259 def test_stub_run_records_no_checkpoint(self):CODE
LOW…h/benchmarks/zh_short_commands/tests/test_benchmark.py96 def test_only_the_none_label_is_not_a_command(self):CODE
LOW…h/benchmarks/zh_short_commands/tests/test_benchmark.py107 def test_every_config_builds_one_pair_per_text(self):CODE
LOW…h/benchmarks/zh_short_commands/tests/test_benchmark.py119 def test_noul_plain_omits_criteria_and_the_higher_rungs_add_them(self):CODE
LOW…h/benchmarks/zh_short_commands/tests/test_benchmark.py127 def test_scenario_rungs_prepend_the_scenario(self):CODE
LOW…h/benchmarks/zh_short_commands/tests/test_benchmark.py204 def test_every_case_is_recorded_once_per_role(self):CODE
LOW…h/benchmarks/zh_short_commands/tests/test_benchmark.py208 def test_probabilities_are_finite_and_normalised(self):CODE
LOW…h/benchmarks/zh_short_commands/tests/test_benchmark.py220 def test_every_record_carries_the_request_it_was_asked(self):CODE
LOW⚡…earch/benchmarks/zh_short_commands/tests/test_audit.py36 def test_manifest_hashes_pin_the_current_files(self):CODE
LOW⚡…earch/benchmarks/zh_short_commands/tests/test_audit.py43 def test_manifest_counts_match_the_cases(self):CODE
LOW⚡…earch/benchmarks/zh_short_commands/tests/test_audit.py52 def test_manifest_freezes_the_ladder_this_file_builds(self):CODE
LOW⚡…earch/benchmarks/zh_short_commands/tests/test_audit.py58 def test_archive_audits_clean(self):CODE
LOW⚡…earch/benchmarks/zh_short_commands/tests/test_audit.py64 def test_archive_pins_the_weights(self):CODE
LOW⚡…earch/benchmarks/zh_short_commands/tests/test_audit.py70 def test_stub_archive_has_no_weight_to_pin(self):CODE
LOW⚡…earch/benchmarks/zh_short_commands/tests/test_audit.py136 def test_probability_out_of_range(self):CODE
LOW⚡…earch/benchmarks/zh_short_commands/tests/test_audit.py141 def test_prediction_that_does_not_follow_from_the_probability(self):CODE
LOW⚡…earch/benchmarks/zh_short_commands/tests/test_audit.py147 def test_noul_confidence_that_is_not_the_maximum(self):CODE
LOW⚡…earch/benchmarks/zh_short_commands/tests/test_audit.py152 def test_protocol_from_a_different_prompt_revision(self):CODE
LOW⚡…earch/benchmarks/zh_short_commands/tests/test_audit.py172 def test_relabelled_dimension(self):CODE
LOW⚡…earch/benchmarks/zh_short_commands/tests/test_audit.py177 def test_summary_that_disagrees_with_its_report(self):CODE
LOW⚡…earch/benchmarks/zh_short_commands/tests/test_audit.py182 def test_unexplained_report_field(self):CODE
LOW⚡…earch/benchmarks/zh_short_commands/tests/test_audit.py187 def test_a_ladder_other_than_the_frozen_one(self):CODE
LOW⚡…earch/benchmarks/zh_short_commands/tests/test_audit.py192 def test_missing_weights_for_a_real_run(self):CODE
LOW⚡…earch/benchmarks/zh_short_commands/tests/test_audit.py232 def test_a_confidence_of_exactly_one_lands_in_the_last_bin(self):CODE
LOW⚡…earch/benchmarks/zh_short_commands/tests/test_audit.py236 def test_a_confidence_of_exactly_zero_is_counted(self):CODE
LOW⚡…earch/benchmarks/zh_short_commands/tests/test_audit.py240 def test_summarise_rounding_matches(self):CODE
862 more matches not shown…
Decorative Section Separators121 hits · 364 pts
SeverityFileLineSnippetContext
MEDIUMtests/test_predict_long.py1183# ---------------------------------------------------------------------------------------------COMMENT
MEDIUMlaya-dotnet/tools/bench_python.py184 # ── cold latency ─────────────────────────────────────────────────────────COMMENT
MEDIUMlaya-dotnet/tools/bench_python.py196 # ── warm latency ──────────────────────────────────────────────────────────COMMENT
MEDIUMlaya-dotnet/tools/bench_python.py223 # ── peak memory ───────────────────────────────────────────────────────────COMMENT
MEDIUMlaya-dotnet/tools/bench_python.py232 # ── answer snapshot ───────────────────────────────────────────────────────COMMENT
MEDIUM⚡laya-dotnet/tools/export_onnx.py652# ─── ONNX graph signature check ─────────────────────────────────────────────COMMENT
MEDIUMlaya-dotnet/tools/export_onnx.py63# ─── model catalogue ────────────────────────────────────────────────────────COMMENT
MEDIUMlaya-dotnet/tools/export_onnx.py169# ─── fixture cases ──────────────────────────────────────────────────────────COMMENT
MEDIUMlaya-dotnet/tools/export_onnx.py308# ─── ONNX export ────────────────────────────────────────────────────────────COMMENT
MEDIUMlaya-dotnet/tools/export_onnx.py435# ─── tokenizer copy ─────────────────────────────────────────────────────────COMMENT
MEDIUMlaya-dotnet/tools/export_onnx.py461# ─── ONNX parity verification ────────────────────────────────────────────────COMMENT
MEDIUMlaya-dotnet/tools/export_onnx.py723# ─── main per-model export ───────────────────────────────────────────────────COMMENT
MEDIUMlaya-dotnet/tools/export_onnx.py823# ─── graph signature check against reference ────────────────────────────────COMMENT
MEDIUMlaya-dotnet/tools/export_onnx.py839# ─── CLI ────────────────────────────────────────────────────────────────────COMMENT
MEDIUMlaya-dotnet/tests/Laya.Tests/LayaEmailTests.cs39 // ── a pure footer is still removed ───────────────────────────────────────COMMENT
MEDIUMlaya-dotnet/tests/Laya.Tests/LayaEmailTests.cs56 // ── unrelated cleaning is unchanged ──────────────────────────────────────COMMENT
MEDIUM⚡laya-dotnet/tests/Laya.Tests/LayaEmailTests.cs96 // ── email_state ───────────────────────────────────────────────────────────COMMENT
MEDIUMlaya-dotnet/tests/Laya.Tests/LayaEmailTests.cs128 // ── every recorded probe entry ────────────────────────────────────────────COMMENT
MEDIUMlaya-dotnet/tests/Laya.Tests/PredictParityTests.cs23 // ── cross-checkpoint theory data ──────────────────────────────────────────COMMENT
MEDIUMlaya-dotnet/tests/Laya.Tests/PredictParityTests.cs60 // ── theory tests ──────────────────────────────────────────────────────────COMMENT
MEDIUMlaya-dotnet/tests/Laya.Tests/PredictParityTests.cs196 // ── fact tests (multilingual checkpoint) ─────────────────────────────────COMMENT
MEDIUMlaya-dotnet/tests/Laya.Tests/PredictParityTests.cs237 // ── helpers ───────────────────────────────────────────────────────────────COMMENT
MEDIUMlaya-dotnet/tests/Laya.Tests/LanguageDetectionTests.cs12 // ── script detection ─────────────────────────────────────────────────────COMMENT
MEDIUMlaya-dotnet/tests/Laya.Tests/LanguageDetectionTests.cs47 // ── is_english / analyse ─────────────────────────────────────────────────COMMENT
MEDIUMlaya-dotnet/tests/Laya.Tests/LanguageDetectionTests.cs105 // ── Latin language guess ─────────────────────────────────────────────────COMMENT
MEDIUMlaya-dotnet/tests/Laya.Tests/LanguageDetectionTests.cs122 // ── state flattening ──────────────────────────────────────────────────────COMMENT
MEDIUMlaya-dotnet/tests/Laya.Tests/LanguageDetectionTests.cs140 // ── script profile ────────────────────────────────────────────────────────COMMENT
MEDIUMlaya-dotnet/tests/Laya.Tests/LanguageDetectionTests.cs151 // ── plain-ASCII Romance (#172): accents stripped must still route non-English ──────────────COMMENT
MEDIUMlaya-dotnet/tests/Laya.Tests/LanguageDetectionTests.cs151 // ── plain-ASCII Romance (#172): accents stripped must still route non-English ──────────────COMMENT
MEDIUMlaya-dotnet/tests/Laya.Tests/LanguageDetectionTests.cs198 // ── every recorded probe entry ───────────────────────────────────────────COMMENT
MEDIUMlaya-dotnet/tests/Laya.Tests/ShapeSweepTests.cs76 // ── multilingual-only shape tests ─────────────────────────────────────────COMMENT
MEDIUMlaya-dotnet/tests/Laya.Tests/ShapeSweepTests.cs150 // ── all-checkpoint shape test ─────────────────────────────────────────────COMMENT
MEDIUMlaya-dotnet/tests/Laya.Tests/ShapeSweepTests.cs168 // ── split-layout shape test ────────────────────────────────────────────────COMMENT
MEDIUMlaya-dotnet/tests/Laya.Tests/ShapeSweepTests.cs228 // ── helpers ───────────────────────────────────────────────────────────────COMMENT
MEDIUMlaya-dotnet/tests/Laya.Tests/PythonJsonTests.cs69 // ── State(null) ───────────────────────────────────────────────────────COMMENT
MEDIUMlaya-dotnet/tests/Laya.Tests/PythonJsonTests.cs79 // ── Repr-specific unit tests ──────────────────────────────────────────COMMENT
MEDIUMlaya-dotnet/tests/Laya.Tests/PythonJsonTests.cs108 // ── Escaping unit tests ───────────────────────────────────────────────COMMENT
MEDIUMlaya-dotnet/tests/Laya.Tests/PresetsTests.cs5 // ── Triage ───────────────────────────────────────────────────────────────COMMENT
MEDIUMlaya-dotnet/tests/Laya.Tests/PresetsTests.cs69 // ── Email ─────────────────────────────────────────────────────────────────COMMENT
MEDIUMlaya-dotnet/tests/Laya.Tests/PresetsTests.cs164 // ── Guard ─────────────────────────────────────────────────────────────────COMMENT
MEDIUMlaya-dotnet/tests/Laya.Tests/PresetsTests.cs231 // ── Moderation ────────────────────────────────────────────────────────────COMMENT
MEDIUMlaya-dotnet/tests/Laya.Tests/PresetsTests.cs285 // ── Router ────────────────────────────────────────────────────────────────COMMENT
MEDIUMlaya-dotnet/tests/Laya.Tests/GoldenData.cs44 // ── availability ──────────────────────────────────────────────────────────COMMENT
MEDIUMlaya-dotnet/tests/Laya.Tests/GoldenData.cs62 // ── data ──────────────────────────────────────────────────────────────────COMMENT
MEDIUMlaya-dotnet/tests/Laya.Tests/GoldenData.cs91 // ── helpers ───────────────────────────────────────────────────────────────COMMENT
MEDIUMlaya-dotnet/tests/Laya.Tests/GoldenData.cs137 // ── backward-compat shims pointing at the multilingual checkpoint ─────────COMMENT
MEDIUMlaya-dotnet/tests/Laya.Tests/GoldenData.cs168 // ── pure utilities (checkpoint-independent) ────────────────────────────────COMMENT
MEDIUMlaya-dotnet/tests/Laya.Tests/LayaRouterTests.cs32 // ── workflow signatures ───────────────────────────────────────────────────COMMENT
MEDIUMlaya-dotnet/tests/Laya.Tests/LayaRouterTests.cs73 // ── name normalisation ────────────────────────────────────────────────────COMMENT
MEDIUMlaya-dotnet/tests/Laya.Tests/LayaRouterTests.cs91 // ── routing decisions ─────────────────────────────────────────────────────COMMENT
MEDIUMlaya-dotnet/tests/Laya.Tests/LayaRouterTests.cs190 // ── LRU bookkeeping ───────────────────────────────────────────────────────COMMENT
MEDIUMlaya-dotnet/tests/Laya.Tests/LayaRouterTests.cs234 // ── preload ───────────────────────────────────────────────────────────────COMMENT
MEDIUMlaya-dotnet/tests/Laya.Tests/LayaRouterTests.cs268 // ── attach ────────────────────────────────────────────────────────────────COMMENT
MEDIUMlaya-dotnet/tests/Laya.Tests/LayaRouterTests.cs359 // ── thread safety (Python issue #95) ─────────────────────────────────────COMMENT
MEDIUMlaya-dotnet/tests/Laya.Tests/LayaRouterTests.cs359 // ── thread safety (Python issue #95) ─────────────────────────────────────COMMENT
MEDIUMlaya-dotnet/tests/Laya.Tests/LayaRouterTests.cs391 // ── Predict populates Routing ─────────────────────────────────────────────COMMENT
MEDIUMlaya-dotnet/tests/Laya.Tests/LayaRouterTests.cs411 // ── every recorded probe entry ────────────────────────────────────────────COMMENT
MEDIUMlaya-dotnet/tests/Laya.Tests/TestArtifacts.cs57 // ── split layout (laya-ts exporter output) ─────────────────────────────────COMMENT
MEDIUMlaya-dotnet/tests/Laya.Tests/TestArtifacts.cs83 // ── resolution ─────────────────────────────────────────────────────────────COMMENT
MEDIUMlaya-dotnet/tests/Laya.Tests/TestArtifacts.cs180 // ── backward-compat shims pointing at the multilingual checkpoint ─────────COMMENT
61 more matches not shown…
Excessive Try-Catch Wrapping235 hits · 252 pts
SeverityFileLineSnippetContext
LOWresearch/evals/shouted_bar_corpora.py187 except Exception: # offline: the same revision, already in the local cacheCODE
LOWresearch/benchmarks/idle_unload/benchmark.py120 except Exception as exc:CODE
LOWresearch/benchmarks/feishu_zh/run.py126 except Exception as e:CODE
LOWresearch/benchmarks/feishu_zh/run.py141 except Exception as e:CODE
LOWresearch/benchmarks/zh_short_commands/run.py180 except Exception: # pragma: no cover - provenance onlyCODE
LOWresearch/benchmarks/zh_short_commands/run.py200 except Exception as exc: # pragma: no cover - provenance onlyCODE
LOWresearch/scripts/build_benchmark_nb.py136 except Exception as e:CODE
LOWresearch/scripts/build_benchmark_nb.py176 except Exception:CODE
LOWresearch/scripts/build_benchmark_nb.py336 except Exception:CODE
LOWresearch/scripts/build_benchmark_nb.py359except Exception as e:CODE
LOWresearch/scripts/build_benchmark_nb.py384 except Exception as e:CODE
LOWresearch/scripts/build_benchmark_nb.py401 except Exception as e:CODE
LOWresearch/scripts/build_benchmark_nb.py425except Exception as e: print(" FAIL sst5:", str(e)[:90])CODE
LOWresearch/scripts/build_benchmark_nb.py439except Exception as e: print(" FAIL emotion:", str(e)[:90])CODE
LOWresearch/scripts/build_benchmark_nb.py451except Exception as e: print(" FAIL prompt_injections:", str(e)[:90])CODE
LOWresearch/scripts/build_benchmark_nb.py465except Exception as e: print(" FAIL banking77:", str(e)[:90])CODE
LOWresearch/scripts/build_benchmark_nb.py479except Exception as e: print(" FAIL ag_news:", str(e)[:90])CODE
LOWresearch/scripts/build_benchmark_nb.py490except Exception as e: print(" FAIL boolq:", str(e)[:90])CODE
LOWresearch/scripts/build_benchmark_nb.py566 except Exception as e:CODE
LOWresearch/scripts/build_benchmark_nb.py817except Exception as e:CODE
LOWresearch/scripts/bench_local.py67 except Exception:CODE
LOWresearch/scripts/bench_local.py184 except Exception as e:CODE
LOWresearch/scripts/bench_local.py235 except Exception:CODE
LOWresearch/scripts/bench_local.py327 except Exception:CODE
LOWresearch/scripts/bench_nvidia_capacity.py179 except Exception as e:CODE
LOWresearch/scripts/bench_nvidia_capacity.py216 except Exception:CODE
LOWresearch/scripts/bench_nvidia_capacity.py274 except Exception:CODE
LOWresearch/scripts/bench_apps.py95 except Exception as e:CODE
LOWresearch/scripts/bench_apps.py111 except Exception as e:CODE
LOWresearch/scripts/bench_apps.py127 except Exception as e:CODE
LOWresearch/scripts/bench_apps.py156 except Exception as e:CODE
LOWresearch/scripts/bench_apps.py169 except Exception as e:CODE
LOWresearch/scripts/bench_apps.py186 except Exception as e:CODE
LOWresearch/scripts/bench_apps.py218 except Exception as e:CODE
LOWresearch/scripts/bench_apps.py240 except Exception as e:CODE
LOWresearch/scripts/bench_apps.py269 except Exception as e:CODE
LOWresearch/scripts/bench_apps.py298 except Exception as e:CODE
LOWresearch/scripts/step4_benchmark/kaggle/step4_gate_r.py54 except Exception:CODE
LOWresearch/scripts/step4_benchmark/kaggle/step4_gate_r.py120 except Exception as _e: # noqa: BLE001 - a missing optional row must not kill the runCODE
LOWresearch/scripts/step4_benchmark/kaggle/build_gate_d.py79 except Exception:CODE
LOWresearch/scripts/step4_benchmark/kaggle/build_kernel.py78 except Exception:STRING
LOWresearch/scripts/step4_benchmark/kaggle/build_kernel.py144 except Exception as _e: # noqa: BLE001 - a missing optional row must not kill the runSTRING
LOW…/scripts/step4_benchmark/kaggle/gate_d/step4_gate_d.py57 except Exception:CODE
LOWresearch/eval/metamorphic.py464 except Exception as exc:CODE
LOWresearch/eval/metamorphic.py689 except Exception as exc:CODE
LOWresearch/eval/laya_eval.py105 except Exception as exc: # pragma: no cover - network pathCODE
LOWresearch/eval/laya_eval.py317 except Exception as exc:CODE
LOWresearch/eval/laya_eval.py357 except Exception as exc:CODE
LOWlaya/train_cli.py233 except Exception as err:CODE
LOWlaya/hooks.py540 except Exception as exc: # noqa: BLE001 -- policy depends on raise_errorsCODE
MEDIUMlaya/hooks.py496def runner():CODE
LOWlaya/evals.py263 except Exception: # pragma: no cover - identity onlyCODE
LOWlaya/evals.py1089 except Exception as exc: # noqa: BLE001 -- honoured by on_errorCODE
MEDIUMlaya/serve.py1073def _run_inference(fn):CODE
LOWlaya/serve.py1098 except Exception: # noqa: BLE001 -- a failed unload must not stop future attemptsCODE
LOWlaya/serve.py1305 except Exception: # noqa: BLE001 -- never leak paths/weights/OOM text to clientsCODE
LOWlaya/serve.py1446 except Exception: # noqa: BLE001CODE
LOWlaya/agent.py107 except Exception as e:CODE
LOWlaya/agent.py416 except Exception:CODE
LOWlaya/agent.py691 except Exception:CODE
175 more matches not shown…
Empty Error Handlers42 hits · 210 pts
SeverityFileLineSnippetContext
HIGHresearch/benchmarks/zh_short_commands/run.py180except Exception: # pragma: no cover - provenance onlyCODE
HIGHresearch/scripts/bench_local.py235except Exception:CODE
HIGHresearch/scripts/bench_local.py327except Exception:CODE
HIGHlaya/serve.py967except ValueError:CODE
HIGHlaya/serve.py1120except asyncio.CancelledError:CODE
HIGHlaya/serve.py1467except OSError:CODE
HIGHlaya/agent.py691except Exception:CODE
HIGHlaya/agent.py1944except Exception:CODE
HIGHlaya/agent.py734except (TypeError, ValueError):CODE
HIGHlaya/agent.py2029except FileNotFoundError:CODE
HIGHlaya/agent.py98except OSError:CODE
HIGHlaya/agent.py2037except OSError:CODE
HIGHlaya/agent.py104except OSError:CODE
HIGHlaya/calibrate.py572except (TypeError, ValueError):CODE
HIGHlaya/router.py1031except Exception:CODE
HIGHlaya/router.py1111except Exception:CODE
HIGHlaya/router.py1142except Exception:CODE
HIGHlaya/onnx_agent.py210except (TypeError, ValueError):CODE
HIGHlaya/backends/compile.py55except Exception:CODE
HIGHlaya/mcp/device.py121except Exception:CODE
HIGHlaya/mcp/device.py139except Exception:CODE
HIGHlaya/mcp/tools.py662except (AttributeError, TypeError):CODE
HIGHlaya/mcp/tools.py911except Exception:CODE
HIGHlaya/integrations/crewai.py398except Exception:CODE
HIGHlaya/integrations/crewai.py420except Exception:CODE
HIGHdocker/check_torch.py64except PackageNotFoundError:CODE
HIGHtests/test_state_budget.py806except _Skip:CODE
HIGHtests/test_mcp_local_e2e.py88except Exception:CODE
HIGHtests/test_router.py1795except TypeError:CODE
HIGHtests/test_router.py1871except Exception: # noqa: BLE001CODE
HIGHtests/test_router.py1853except ValueError: # unregistered at that instantCODE
HIGHtests/test_calibrate.py352except OSError:CODE
HIGHtests/test_compile.py140except RuntimeError:CODE
HIGHtests/test_compile.py367except RuntimeError:CODE
HIGHtests/test_agent_gate.py193except RuntimeError:CODE
HIGHtests/test_agent_gate.py201except RuntimeError:CODE
HIGHtests/test_cli.py133except SystemExit:CODE
HIGHlaya-dotnet/tools/export_onnx.py631except Exception:CODE
HIGHlaya-dotnet/tools/export_onnx.py780except Exception:CODE
HIGHlaya-dotnet/tools/export_onnx.py790except Exception:CODE
HIGHlaya-dotnet/tools/export_onnx.py368except (AttributeError, io.UnsupportedOperation):CODE
HIGHlaya-dotnet/tools/export_onnx.py642except Exception:CODE
Over-Commented Block198 hits · 193 pts
SeverityFileLineSnippetContext
LOWcompose.http.yaml1# Serve Laya over HTTP from the same image, so a Jev client can point its `baseUrl` at it:COMMENT
LOWcompose.http.yaml61 # third checkpoint routing can choose on demand, and a cap below it reloads one per switch.COMMENT
LOWcompose.modelscope.yaml1# Bake the checkpoint from ModelScope (https://modelscope.cn) into the image, so neither the buildCOMMENT
LOWcompose.modelscope.yaml21# For NVIDIA, append -f compose.cuda.yaml; it repeats the build args and the device reservation forCOMMENT
LOWcompose.example.yml21 # Optional token: enable this setting and its read-only mount below.COMMENT
LOWlaya/lang.py201 if sum(w in words for words in _STOP.values()) > 1}COMMENT
LOWlaya/lang.py221# ... van" is ordinary Dutch, and deduping it moved 14 of those rows while rescuing no English one.COMMENT
LOWlaya/lang.py501 return latin_profile(text)["language"]COMMENT
LOWlaya/lang.py521# ...but a line of nothing but acronyms has no lowercase either, so it skips the blanking aboveCOMMENT
LOWlaya/lang.py541# word >= 5 only 10,854 99.03% 431 141 <- addresses standCOMMENT
LOWlaya/lang.py561_SHOUTED_MIN_STOPWORD = 4COMMENT
LOWlaya/lang.py681 "is_english": True, "language_undecided": True, "diacritic_rate": 0.0,COMMENT
LOWlaya/lang.py701 # letters and was being sent to the English checkpoint on the strength of disbelieving `MONCOMMENT
LOWlaya/lang.py721 # that carry the sentence, so when the shouted words were the ones that named theCOMMENT
LOWlaya/lang.py761 for line in leaf.split("\n"):COMMENT
LOWlaya/lang.py781 # non-English letters and that no stopword list can name. `WIE SPÄT IST ES IN KÖLN`COMMENT
LOWlaya/__init__.py1"""Laya: Fast, non-autoregressive System 1 decision engine with calibrated probabilities."""COMMENT
LOWlaya/evals.py261 try:COMMENT
LOWlaya/serve.py81# A failed inference is reported to the client as a fixed 500 so nothing about paths,COMMENT
LOWlaya/serve.py101# What a row COSTS is its width, and `max_len` is a request field capped only byCOMMENT
LOWlaya/serve.py121# checkpoint's own `max_len` without loading it, so this is the shipped english value; multilingualCOMMENT
LOWlaya/serve.py141# open; the rest of the payload names checkpoints, revision SHAs and host device state, so itCOMMENT
LOWlaya/serve.py781 if over:COMMENT
LOWlaya/serve.py1041 router = build_router()COMMENT
LOWlaya/serve.py1341 # body rather than one state because that is what the batch carries: `predict_batch`COMMENT
LOWlaya/email.py21 # "Em resposta ao que você escreveu:" is body text; a client's attribution always carries a dateCOMMENT
LOWlaya/email.py41]COMMENT
LOWlaya/agent.py741 "laya: this checkpoint ships invalid temperatures or values outside [%g, %g]; "COMMENT
LOWlaya/agent.py1021 # over: a label is option text AND the answer key, and a null one can beCOMMENT
LOWlaya/agent.py1721 internal = {qid: self._to_internal(questions[qid]) for qid in ids}COMMENT
LOWlaya/agent.py1741 # "Fits" is the room the questions leave, not the default window: `system_one` reads aCOMMENT
LOWlaya/agent.py1841 # deciding span; the single most-confident window preserves a localized signal.COMMENT
LOWlaya/router.py341 if _takes_lang(scan):COMMENT
LOWlaya/router.py361 "laya: Router.predict_long: %s.predict_long has no `lang` parameter, so the %s "COMMENT
LOWlaya/router.py761 # the `_loading` registry, and lifecycle bookkeeping. Never held during expensiveCOMMENT
LOWlaya/onnx_agent.py461 if not isinstance(questions, dict):COMMENT
LOWlaya/onnx_agent.py581 # choice / score: the most-confident window wins, preserving a localized signalCOMMENT
LOWtests/test_compose_env.py221for name in names:COMMENT
LOWtests/test_compose_env.py241# A file may still deviate, and the test for that is *where the deviation is said*: the entryCOMMENT
LOWtests/test_compose_env.py401 "the omission has to be explained where a reader meets it")COMMENT
LOWtests/test_compose_env.py421COMMENT
LOWtests/test_doc_tables.py201 rounded(max(v["d_fast_fp32"] for v in fp16_rows)), "0.009")COMMENT
LOWtests/test_download.py341 def test_onnx_agent_accepts_every_hub_option_the_agent_does(self):COMMENT
LOWtests/test_onnx_lang_parity.py121 p_de["billing"] < p_base["billing"])COMMENT
LOWtests/test_packaging.py241check("ci/wires every non-exempt test suite", untested_suites, [])COMMENT
LOWtests/test_packaging.py581undocumented = [n for n in accepted if "`%s`" % n not in described]COMMENT
LOWtests/test_mcp_local_e2e.py381 "questions": ticket["questions"],COMMENT
LOWtests/test_load_errors.py181 _outcome = excCOMMENT
LOWtests/test_load_errors.py241# the tuple `(repo, subfolder)` when the directory is missing: `TypeError: expected str, bytesCOMMENT
LOWtests/test_calibration_persistence.py241 # #642 fixed this for the Kaggle notebook, which owned a fit of its own. TheCOMMENT
LOWtests/test_router.py201check("route/dict german after long english note", _r_de.route(_de_long, {}).model, "multilingual")COMMENT
LOWtests/test_router.py1521unloaded_ok, not_res_after, is_new_agent, res_final = _test_unload_reload_sequence_no_resurrection()COMMENT
LOWtests/test_env_docs.py621COMMENT
LOWtests/test_shortlist.py781COMMENT
LOWtests/test_predict_long.py941COMMENT
LOWtests/test_structured.py221# entropy: `laya/common.py` calls it "not calibrated: it is not what temperature scaling fits andCOMMENT
LOWtests/test_example_server_limits.py281 ("an ordinary request", {"body": "billed twice"}, one),COMMENT
LOWtests/test_example_server_limits.py441 "%r vs %r" % (fresh._router_kwargs(fresh._CFG), build_router().max_loaded))COMMENT
LOWtests/test_runtime_fixes.py381check("oom-fallback/plain CPU infer stays 0", cpu_disabled.cpu_fallback_count, 0)COMMENT
LOWtests/test_lazy_import.py81COMMENT
138 more matches not shown…
Hallucination Indicators16 hits · 160 pts
SeverityFileLineSnippetContext
CRITICAL…/com/convaiinnovations/laya/QuestionRenderingTest.java56 org.junit.jupiter.api.Assertions.assertTrue(rendered.contains("a: {\"d\": 1}"), rendered.toString());CODE
CRITICAL…/com/convaiinnovations/laya/QuestionRenderingTest.java57 org.junit.jupiter.api.Assertions.assertTrue(rendered.contains("b: [1, 2]"), rendered.toString());CODE
CRITICAL…/com/convaiinnovations/laya/QuestionRenderingTest.java156 org.junit.jupiter.api.Assertions.assertTrue(refused.getMessage().contains("level 1"),CODE
CRITICAL…/com/convaiinnovations/laya/QuestionRenderingTest.java180 org.junit.jupiter.api.Assertions.assertThrows(IllegalArgumentException.class,CODE
CRITICAL…/com/convaiinnovations/laya/QuestionRenderingTest.java201 org.junit.jupiter.api.Assertions.assertThrows(IllegalArgumentException.class,CODE
CRITICAL…java/com/convaiinnovations/laya/AgentBatchingTest.java140 org.junit.jupiter.api.Assertions.assertNotNull(collapsed,CODE
CRITICAL…st/java/com/convaiinnovations/laya/AgentHooksTest.java83 com.convaiinnovations.laya.tokenizer.Tokenizer.fromModelDirectory(root),CODE
CRITICAL…st/java/com/convaiinnovations/laya/AgentHooksTest.java84 com.convaiinnovations.laya.config.AgentConfig.fromModelDirectory(root),CODE
CRITICAL…com/convaiinnovations/laya/config/AgentConfigTest.java73 org.junit.jupiter.api.Assertions.assertThrows(CODE
CRITICAL…com/convaiinnovations/laya/config/AgentConfigTest.java78 org.junit.jupiter.api.Assertions.assertThrows(CODE
CRITICAL…va/com/convaiinnovations/laya/json/PythonJsonTest.java42 org.junit.jupiter.api.Assertions.assertTrue(checked > 400,CODE
CRITICAL…va/com/convaiinnovations/laya/json/PythonJsonTest.java47 org.junit.jupiter.api.Assertions.assertTrue(covered.contains(discriminating)CODE
CRITICAL…va/com/convaiinnovations/laya/json/PythonJsonTest.java82 org.junit.jupiter.api.Assertions.assertThrows(Json.JsonException.class,CODE
CRITICAL…va/com/convaiinnovations/laya/json/PythonJsonTest.java106 org.junit.jupiter.api.Assertions.assertThrows(Json.JsonException.class,CODE
CRITICAL…va/com/convaiinnovations/laya/json/PythonJsonTest.java119 Json.JsonException cyclic = org.junit.jupiter.api.Assertions.assertThrows(CODE
CRITICAL…va/com/convaiinnovations/laya/json/PythonJsonTest.java125 org.junit.jupiter.api.Assertions.assertThrows(Json.JsonException.class,CODE
Deep Nesting143 hits · 123 pts
SeverityFileLineSnippetContext
LOWresearch/evals/shouted_bar_corpora.py218CODE
LOWresearch/benchmarks/tournament/benchmark.py89CODE
LOWresearch/benchmarks/feishu_zh/run.py26CODE
LOWresearch/benchmarks/feishu_zh/render_cards.py38CODE
LOWresearch/benchmarks/feishu_zh/tests/test_audit.py35CODE
LOWresearch/benchmarks/es_phone_turns/train/build_train.py175CODE
LOWresearch/benchmarks/zh_short_commands/audit.py121CODE
LOWresearch/benchmarks/zh_short_commands/audit.py202CODE
LOWresearch/scripts/bench_position_sensitivity.py83CODE
LOWresearch/scripts/bench_local.py225CODE
LOWresearch/scripts/bench_long_context.py53CODE
LOWresearch/scripts/bench_question_token_reuse.py72CODE
LOWresearch/scripts/bench_length_batching.py60CODE
LOWresearch/scripts/bench_nvidia_capacity.py149CODE
LOWresearch/scripts/bench_nvidia_capacity.py297CODE
LOWresearch/scripts/bench_nvidia_capacity.py163CODE
LOWresearch/scripts/finetune_single_device.py32CODE
LOWresearch/scripts/step4_benchmark/analyze_gate_d.py61CODE
LOWresearch/scripts/step4_benchmark/evaluate.py52CODE
LOWlaya/lang.py433CODE
LOWlaya/lang.py665CODE
LOWlaya/lang.py749CODE
LOWlaya/hooks.py512CODE
LOWlaya/hooks.py413CODE
LOWlaya/_agreement.py206CODE
LOWlaya/confidence.py247CODE
LOWlaya/evals.py925CODE
LOWlaya/serve.py182CODE
LOWlaya/serve.py693CODE
LOWlaya/serve.py800CODE
LOWlaya/serve.py907CODE
LOWlaya/cli.py199CODE
LOWlaya/cli.py317CODE
LOWlaya/structured.py261CODE
LOWlaya/agent.py61CODE
LOWlaya/agent.py485CODE
LOWlaya/agent.py968CODE
LOWlaya/agent.py1207CODE
LOWlaya/agent.py1409CODE
LOWlaya/agent.py1600CODE
LOWlaya/train.py196CODE
LOWlaya/train.py317CODE
LOWlaya/train.py649CODE
LOWlaya/train.py754CODE
LOWlaya/train.py807CODE
LOWlaya/train.py989CODE
LOWlaya/calibrate.py607CODE
LOWlaya/router.py791CODE
LOWlaya/router.py914CODE
LOWlaya/router.py1007CODE
LOWlaya/router.py1090CODE
LOWlaya/router.py1117CODE
LOWlaya/router.py1203CODE
LOWlaya/router.py1302CODE
LOWlaya/router.py1590CODE
LOWlaya/router.py1828CODE
LOWlaya/onnx_agent.py293CODE
LOWlaya/onnx_agent.py398CODE
LOWlaya/onnx_agent.py736CODE
LOWlaya/tl_kernels.py182CODE
83 more matches not shown…
Unused Imports98 hits · 96 pts
SeverityFileLineSnippetContext
LOWresearch/evals/act_head_eval.py69CODE
LOWresearch/evals/massive_route_comparison.py16CODE
LOWresearch/evals/check_regression.py9CODE
LOWresearch/evals/shouted_bar_corpora.py30CODE
LOWresearch/benchmarks/zh_short_commands/run.py28CODE
LOWresearch/benchmarks/zh_short_commands/run.py28CODE
LOWresearch/scripts/bench_apps.py25CODE
LOWresearch/scripts/bench_apps.py35CODE
LOWresearch/scripts/bench_apps.py36CODE
LOWresearch/scripts/step4_benchmark/smoke_test.py13CODE
LOWresearch/scripts/step4_benchmark/run_benchmark.py24CODE
LOWresearch/scripts/step4_benchmark/metrics.py19CODE
LOWresearch/scripts/step4_benchmark/analyze_gate_d.py19CODE
LOWresearch/scripts/step4_benchmark/prepare_data.py19CODE
LOWresearch/scripts/step4_benchmark/order_flip.py24CODE
LOWresearch/scripts/step4_benchmark/evaluate.py28CODE
LOWresearch/scripts/step4_benchmark/kaggle/build_gate_d.py14CODE
LOWresearch/scripts/step4_benchmark/kaggle/build_kernel.py17CODE
LOWresearch/scripts/step4_benchmark/kaggle/build_kernel.py21CODE
LOWresearch/eval/metamorphic.py6CODE
LOWresearch/eval/presentation_checks.py58CODE
LOWresearch/eval/laya_eval.py36CODE
LOWlaya/train_cli.py11CODE
LOWlaya/hooks.py9CODE
LOWlaya/evals_shortlist.py7CODE
LOWlaya/__init__.py24CODE
LOWlaya/__init__.py24CODE
LOWlaya/__init__.py25CODE
LOWlaya/__init__.py25CODE
LOWlaya/__init__.py25CODE
LOWlaya/__init__.py25CODE
LOWlaya/__init__.py25CODE
LOWlaya/__init__.py26CODE
LOWlaya/__init__.py27CODE
LOWlaya/__init__.py27CODE
LOWlaya/__init__.py28CODE
LOWlaya/__init__.py28CODE
LOWlaya/__init__.py28CODE
LOWlaya/__init__.py28CODE
LOWlaya/__init__.py28CODE
LOWlaya/__init__.py35CODE
LOWlaya/__init__.py35CODE
LOWlaya/__init__.py35CODE
LOWlaya/__init__.py36CODE
LOWlaya/__init__.py36CODE
LOWlaya/__init__.py36CODE
LOWlaya/evals.py22CODE
LOWlaya/email.py17CODE
LOWlaya/structured.py17CODE
LOWlaya/structured.py84CODE
LOWlaya/evidence.py15CODE
LOWlaya/evals_cli.py12CODE
LOWlaya/backends/__init__.py18CODE
LOWlaya/backends/__init__.py18CODE
LOWlaya/backends/__init__.py36CODE
LOWlaya/mcp/server.py25CODE
LOWlaya/mcp/remote.py22CODE
LOWlaya/mcp/device.py23CODE
LOWlaya/mcp/tools.py3CODE
LOWlaya/integrations/_controls.py14CODE
38 more matches not shown…
AI Slop Vocabulary19 hits · 54 pts
SeverityFileLineSnippetContext
MEDIUMresearch/evals/test_act_head_eval.py185# malformed report can carry them into the harness. They are not measurements: rankingCOMMENT
MEDIUMresearch/eval/test_laya_eval.py107# harness tested `conf > lo` for every bin until the divergence was found, which made itCOMMENT
MEDIUMlaya/evals.py849 # Not introspectable (a C-level or hand-rolled ``__call__``): keep the call this harnessCOMMENT
MEDIUMlaya/evals.py1052 # did the harness issue", which is settled the moment the chunk is grouped, while theCOMMENT
MEDIUMlaya/evals.py1148 # run with no batch form asked for something this harness cannot do.COMMENT
MEDIUMtests/test_doc_tables.py207 # figures with an independently written harness and landed within 0.0003, which is theCOMMENT
MEDIUMtests/test_doc_tables.py311 # divergence interesting rather than a harness change), and the `laya-multilingual` columnCOMMENT
MEDIUMtests/test_evals_onnx.py177 # batching reaches the runner's predict_batch through the harnessCOMMENT
MEDIUMtests/test_evals_onnx.py184 # report is the witness: `sort_by_length_sent` is what the harness put on the wire, not whatCOMMENT
MEDIUMtests/test_evals_api.py64# `evaluate` is the harness entry point callers extend from their own scripts, so its parameterCOMMENT
MEDIUMtests/test_predict_long.py839_q_onnx = {"a": q_many(12)} # 100 options fills this tiny harness entirelyCODE
MEDIUMtests/test_structured.py251# and the eval harness use.COMMENT
LOWtests/test_system_one_lang.py16 # Mock predict_batch to just return the lang it was given so we can test the pass-throughCOMMENT
MEDIUMtests/test_evals.py946# changing. `laya-evals run --batch-size` could bound a pass but not group one, so the harness thatCOMMENT
MEDIUMtests/test_evals.py1052# reads the same numbers at every threshold. What the harness forwards is the *claim* theCOMMENT
MEDIUMtests/test_evals.py1056# thing this harness exists to be.COMMENT
MEDIUMtests/test_evals.py1494 # Why the runner records it itself: the counters under test are the harness's own account, soCOMMENT
MEDIUM.github/workflows/ci.yml146 # The eval harness needs numpy and laya only, so it stays a separate, weight-free job: metricCOMMENT
MEDIUM.github/workflows/ci.yml183 # `test_presentation_checks.py` and `test_act_head_eval.py` guard the eval harness's ownCOMMENT
Modern AI Meta-Vocabulary18 hits · 54 pts
SeverityFileLineSnippetContext
MEDIUMpyproject.toml88# `pip install laya[crewai]` adds the CrewAI multi-agent delegation router (laya.integrations).COMMENT
MEDIUMREADME.md1201# 2. Inline prompt guardrailsCOMMENT
MEDIUMREADME.md1632[Issue #102](https://github.com/NandhaKishorM/laya/issues/102) reports that a top-20 zero-shot shortlist moved a BANKINGCODE
MEDIUMresearch/scripts/bench_apps.py189 # ------------------------------------- 3 & 5. guardrails + moderation (toxic-chat, HELD OUT)COMMENT
MEDIUMresearch/scripts/bench_apps.py221 # ---------------------------------------------------------- 4. RAG passage filterCOMMENT
MEDIUMlaya/mcp/server.py79# Shared usage guardrails for the decision tools. laya_status reports insteadCOMMENT
MEDIUMtests/test_llamaindex.py399# arguments since #530 / #532; this surface forwarded only `model`, so a RAG pipeline choosingCOMMENT
MEDIUMtests/test_question_token_reuse.py78 # The real undecorated method is the same orchestration without reuse.COMMENT
MEDIUMtests/test_batch.py66# --------------------------------------------------------------- orchestration (weight-free fake)COMMENT
MEDIUMtests/test_batch.py349# Fake CUDA device and autocast so this orchestration test also runs in CPU-only CI.COMMENT
MEDIUMtests/test_mcp.py2657 # The guardrails constant is on the decision tools only;COMMENT
MEDIUMlaya-dotnet/README.md401// Prompt guardrailsCOMMENT
MEDIUMlaya-dotnet/README.md437## Hosting: lifetime, threading and dependency injectionCOMMENT
MEDIUMlaya-dotnet/src/Laya/LayaPresets.cs87 /// <summary>Preset questions for real-time LLM input guardrails.</summary>COMMENT
MEDIUMdocs/use-cases.md227## 4. Air-Gapped Edge RAG RouterCOMMENT
MEDIUMdocs/hooks/patterns.md11 - [Guardrails](#guardrails)CODE
MEDIUM⚡examples/llamaindex_quickstart.py62# 1. Sub-35ms Single-Choice RAG Selector (Replaces LLMSingleSelector)COMMENT
MEDIUMlaya-java/README.md397## A state longer than the context windowCOMMENT
Cross-File Repetition9 hits · 45 pts
SeverityFileLineSnippetContext
HIGHlaya/integrations/crewai.py0do not forward bearer credentials across an origin or https downgrade.STRING
HIGHlaya/integrations/llamaindex.py0do not forward bearer credentials across an origin or https downgrade.STRING
HIGHlaya/integrations/langchain.py0do not forward bearer credentials across an origin or https downgrade.STRING
HIGHtests/test_langchain.py0mock agent returning deterministic responses for testing.STRING
HIGHtests/test_llamaindex.py0mock agent returning deterministic responses for testing.STRING
HIGHtests/test_crewai.py0mock agent returning deterministic responses for testing.STRING
HIGHlaya-dotnet/tools/regen_golden.py0refuses a snapshot that is missing files, or whose files arrived empty or unparseable. `snapshot_download` reports succeSTRING
HIGHlaya-dotnet/tools/export_onnx.py0refuses a snapshot that is missing files, or whose files arrived empty or unparseable. `snapshot_download` reports succeSTRING
HIGHlaya-java/scripts/prepare_checkpoint.py0refuses a snapshot that is missing files, or whose files arrived empty or unparseable. `snapshot_download` reports succeSTRING
AI Structural Patterns35 hits · 31 pts
SeverityFileLineSnippetContext
LOWlaya/common.py156CODE
LOWlaya/agent.py2084CODE
LOWlaya/agent.py485CODE
LOWlaya/agent.py1409CODE
LOWlaya/agent.py1600CODE
LOWlaya/agent.py1882CODE
LOWlaya/router.py680CODE
LOWlaya/router.py1175CODE
LOWlaya/router.py1302CODE
LOWlaya/router.py1431CODE
LOWlaya/router.py1590CODE
LOWlaya/onnx_agent.py50CODE
LOWlaya/onnx_agent.py250CODE
LOWlaya/onnx_agent.py293CODE
LOWlaya/onnx_agent.py398CODE
LOWlaya/tl_kernels.py38CODE
LOWlaya/mcp/server.py402CODE
LOWlaya/mcp/remote.py161CODE
LOWlaya/mcp/remote.py194CODE
LOWlaya/mcp/tools.py409CODE
LOWlaya/mcp/tools.py669CODE
LOWlaya/mcp/tools.py854CODE
LOWlaya/integrations/crewai.py209CODE
LOWlaya/integrations/crewai.py251CODE
LOWlaya/integrations/crewai.py432CODE
LOWlaya/integrations/llamaindex.py205CODE
LOWlaya/integrations/llamaindex.py248CODE
LOWlaya/integrations/llamaindex.py394CODE
LOWlaya/integrations/llamaindex.py548CODE
LOWlaya/integrations/langchain.py220CODE
LOWlaya/integrations/langchain.py423CODE
LOWlaya/integrations/langchain.py581CODE
LOWlaya/integrations/langchain.py768CODE
LOWlaya/integrations/langchain.py898CODE
LOWlaya/integrations/langchain.py1065CODE
Fake / Example Data10 hits · 11 pts
SeverityFileLineSnippetContext
LOWlaya-dotnet/tools/routing_cases.py517 ("with_sender", "Refund request", "Please refund invoice 4411.", "user@example.com", True, {}),CODE
LOWlaya-dotnet/tools/routing_cases.py519 ("extra_fields_kept_when_not_none", "Ticket", "Body text.", "user@example.com", True, {"priority": "high", "ticket_iCODE
LOW⚡laya-dotnet/tests/Laya.Tests/LayaEmailTests.cs101 var state = LayaEmail.State("Refund request", "Please refund invoice 4411.", sender: "user@example.com");CODE
LOW⚡laya-dotnet/tests/Laya.Tests/LayaEmailTests.cs102 Assert.Equal("user@example.com", state["from"]);CODE
LOW…otnet/tests/Laya.Tests/golden/routing/email_probe.json453 "sender": "user@example.com",CODE
LOW…otnet/tests/Laya.Tests/golden/routing/email_probe.json459 "from": "user@example.com"CODE
LOW…otnet/tests/Laya.Tests/golden/routing/email_probe.json478 "sender": "user@example.com",CODE
LOW…otnet/tests/Laya.Tests/golden/routing/email_probe.json487 "from": "user@example.com",CODE
LOW…est/java/com/convaiinnovations/laya/ShortlistTest.java195 assertEquals(expectedOptions, Question.choice("placeholder", criteria).renderOptions(),CODE
LOW…rc/main/java/com/convaiinnovations/laya/Shortlist.java189 List<String> optionTexts = Question.choice("placeholder", options).renderOptions();CODE
Modern Structural Boilerplate6 hits · 6 pts
SeverityFileLineSnippetContext
LOWlaya/hooks.py211def set_default_hooks(hooks=None, on_predict_start=None, on_predict_end=None) -> None:CODE
LOWlaya/__init__.py100__all__ = [CODE
LOWlaya/structured.py442__all__ = [CODE
LOWlaya/backends/__init__.py110__all__ = ["Backend", "BackendUnavailable", "NAMES", "auto_policy", "bucket_shape",CODE
LOWlaya/mcp/__init__.py16__all__ = ["device", "server", "tools"]CODE
LOWlaya/integrations/__init__.py23__all__ = [CODE
Structural Annotation Overuse4 hits · 6 pts
SeverityFileLineSnippetContext
LOWlaya-dotnet/MODELS.md27### Step 1: Create a virtual environment and installCOMMENT
LOWlaya-dotnet/MODELS.md98### Step 4: ExportCOMMENT
LOWlaya-dotnet/MODELS.md136### Step 5: Point the SDK at the exportCOMMENT
LOWlaya-dotnet/MODELS.md176### Step 6: Check it worksCOMMENT
Docstring Block Structure1 hit · 5 pts
SeverityFileLineSnippetContext
HIGHlaya/router.py1439Route, then scan every window of the state instead of only its first one. `predict` scores a state from a singlSTRING
Self-Referential Comments1 hit · 3 pts
SeverityFileLineSnippetContext
MEDIUMcompose.http.yaml12# This file is additive. `laya` in compose.yaml still runs the one-shot SDK quickstart andCOMMENT
Redundant / Tautological Comments1 hit · 2 pts
SeverityFileLineSnippetContext
LOWlaya-dotnet/tools/export_onnx.py613 # Check if there is a safetensors file alongside the onnx output (for local-path exports).COMMENT