issues Search Results · language:Dune language:JavaScript language:Python language:JavaScript language:Java language:Java
Filter by
55.1M results
Parent
https://github.com/daKmoR/pull-ops/issues/59
What to build
Implement local prd:auto-advance behavior while preserving PRD Child Coordination semantics and human Child Issue PR
review gates. Dry-run ...
ready-for-agent
daKmoR
error_from_stream_payload() converts error.code with int(error.get( code ) or 0). The non-stream error path preserves
string codes, and API error codes commonly arrive as strings such as BAD_REQUEST; an ...
bug
coding-loop/inspected
coding-loop/managed
saxon134
Parent
https://github.com/daKmoR/pull-ops/issues/59
What to build
Support --until finalized for local issue implementation. The Finalized Run Goal should continue through local PullOps
follow-up operations ...
ready-for-agent
Summary
Two minor polish items in the LoRA adapter eval path (feat/lora-adapter-eval). Neither breaks current metrics — the
adapter is correctly captured for attribution (parsed from sys.argv in wandb-metadata.json ...
enhancement
yidakra
Summary
Evaluator.evaluate in scripts/eval.py (sentence-level XCOMET path, :154-170) splits each document s MT and reference
into sentences and scores them pairwise, but on a sentence-count mismatch it ...
bug
Summary
compute_track2_metrics.py matches each inference output row to its reference by the first 120 characters of the source
text (:233-260):
key = r.get(src_lang, ).strip()[:120]
...
ref_rows = ...
bug
Frequently edits apply to a word, sentence, line, paragraph, or something inside there. When the text is manageable, the
.mb file should include the literal original content so it s more parseable by the ...
dandriscoll
Atualizar os módulos centrais do backend e integrar novas funcionalidades.
Tarefas:
- Atualizar auth.controller.js
- Atualizar user.controller.js
- Atualizar school.controller.js
- Atualizar ...
BackEnd
integração
KauanPedreira
Summary
The internal Track 2 metric scores only the run s own mode dict, whereas the official scorer always measures the proper
terms (and separately the random terms) regardless of which mode was shown ...
bug
Summary
compute_track2_metrics.py:274 computes BLEU with sacrebleu s default tokenizer (13a):
bleu = sacrebleu.corpus_bleu(mt_texts, [ref_texts]).score
The official Track 2 scorer selects the tokenizer ...
bug
Learn how you can use GitHub Issues to plan and track your work.
Save views for sprints, backlogs, teams, or releases. Rank, sort, and filter issues to suit the occasion. The possibilities are endless.Learn more about GitHub IssuesProTip! Restrict your search to the title by using the in:title qualifier.
Learn how you can use GitHub Issues to plan and track your work.
Save views for sprints, backlogs, teams, or releases. Rank, sort, and filter issues to suit the occasion. The possibilities are endless.Learn more about GitHub IssuesProTip! Restrict your search to the title by using the in:title qualifier.