Search results

perf: use Anthropic Message Batches API for eval extraction (50% cost reduction)

performancePerformance issuesPerformance issues

Status: Open.

#83 In hdviettt/agenternal;

· hdviettt opened on Apr 8, 2026

feat: add simple entity index to eval harness for temporal retrieval

evalEvaluation pipelineEvaluation pipeline

Status: Open.

#82 In hdviettt/agenternal;

· hdviettt opened on Apr 8, 2026

feat: multi-query decomposition for compositional questions

enhancementNew feature or requestNew feature or request

Status: Open.

#81 In hdviettt/agenternal;

· hdviettt opened on Apr 8, 2026

feat: add --types filter and --workers flag to LongMemEval harness

evalEvaluation pipelineEvaluation pipeline

Status: Open.

#80 In hdviettt/agenternal;

· hdviettt opened on Apr 8, 2026

eval: targeted LongMemEval-M re-run (temporal + preference only)

evalEvaluation pipelineEvaluation pipeline

Status: Open.

#79 In hdviettt/agenternal;

· hdviettt opened on Apr 8, 2026

test: run mechanism tests to validate recent changes

evalEvaluation pipelineEvaluation pipeline

Status: Open.

#78 In hdviettt/agenternal;

· hdviettt opened on Apr 8, 2026

refactor: consume agenternal-pkg instead of local memory modules

Status: Open.

#77 In hdviettt/agenternal;

· hdviettt opened on Apr 6, 2026

feat: improve temporal reasoning accuracy

Status: Open.

#75 In hdviettt/agenternal;

· bach2035 opened on Apr 6, 2026

NeurIPS: human evaluation — validate LLM judge with annotator agreement

evalEvaluation pipelineEvaluation pipeline

Status: Open.

#29 In hdviettt/agenternal;

· hdviettt opened on Apr 5, 2026