Skip to content

pullrequests Search Results · language:Dune language:JavaScript language:JavaScript language:Python language:Java language:Java

Filter by

184M results  (2 s)

184M results

💡 What: Optimize mean outer product calculation by replacing torch.bmm with (rho.T @ rho) / batch_size. 🎯 Why: Prevent O(B * N^2) intermediate memory allocations and speed up matrix multiplications in ...

Adds a small profile activity marker for repository maintenance tracking.

in this follow up we addressing multiple issues for migration 1) the double m suffix issue from back to back migrations which made subsequent migrations fail 2) tertiary node not being registered on ...

Next is finally here

Orientação (tarefas.orientacao) agora aparece no card de contexto da RAT (iniciar/reabrir) e no card da lista de Tarefas (2 linhas). Antes o técnico só via no detalhe da Tarefa — ficava no escuro durante ...

Summary PM님 현장 확인 — PVD03이 코팅중으로 뜨던 것은 이온소스 전력 6.1KW가 장비가 잘못 보내는 오보값(실제 출력 0)이기 때문이었습니다. 코팅중 판정을 증착 파워(파워A·파워B) 기준으로 한정해, 잘못된 이온소스 값에 흔들리지 않게 했습니다. 원인 / 수정 | | 내용 | | :-- | :-- | | 원인 | 기존 코팅중 판정이 ...

Summary Apply S/R Behavior Classifier to every ATM±3 LTP (14 legs) + add cross-strike contributor #31 to COMPOSITE BIAS. New helper classify_leg_sr_behavior(df_l, ltp) — same state machine as the spot ...

cc @AarushSah @nmayorga7 Summary Add LiteLLM as a model provider, enabling benchmarks to run against 100+ LLM providers (OpenAI, Anthropic, Azure, Bedrock, Vertex, Groq, Ollama, etc.) through a single ...

Seed the benchmark with the first batch of CUA-necessity tasks mined from mypcbench. mypcbench targets a Linux desktop of seeded web apps graded by an LLM judge, so its tasks can t be imported literally; ...

Pull Request.