WekaTrace corpus derived from SemiAnalysis Claude Code proxy traces. Built 2026-06-21 17:49:45 UTC via utils/agentic/build_weka_hf_dataset.py.
Derived from semianalysisai/cc-traces-weka-062126 by applying the 256k per-request cap and preserving the surviving requests' relative timestamps.
max_tokens<=64 AND no tools → SUGGESTION MODE, title-gen, Security Monitor)(timestamp, model, in, out, dur_ms, agent_id)input + output ≤ 256,000 tokens. Applied at request granularity, not conversation. Main-agent and sub-agent inner requests are evaluated independently.t values keep their original relative offsets, including sub-agent overlap. If the first request was filtered, all surviving timestamps are shifted by one uniform offset so the earliest survivor starts at t = 0.traces: 393
main_turns: 28,444
subagent_groups: 1,697
subagent_inner_requests: 39,822
total_model_requests: 68,266
total_input_tokens: 6,891,228,864
total_output_tokens: 58,728,807




python utils/agentic/sample_proxy_traces.py --out '<workdir>/proxy' --sampling top --min-trace-version 7 --max-trace-version 7 --min-requests 20 --require-cli-min 2.1.139 --max-parallel-subagents 10 --exclude-dynamic-workflow-bug
Load in aiperf via:
--public-dataset semianalysis_cc_traces_weka_with_subagents_256k
2 commits
WekaTrace corpus derived from SemiAnalysis Claude Code proxy traces. Built 2026-06-21 17:49:45 UTC via utils/agentic/build_weka_hf_dataset.py.
Derived from semianalysisai/cc-traces-weka-062126 by applying the 256k per-request cap and preserving the surviving requests' relative timestamps.
max_tokens<=64 AND no tools → SUGGESTION MODE, title-gen, Security Monitor)(timestamp, model, in, out, dur_ms, agent_id)input + output ≤ 256,000 tokens. Applied at request granularity, not conversation. Main-agent and sub-agent inner requests are evaluated independently.t values keep their original relative offsets, including sub-agent overlap. If the first request was filtered, all surviving timestamps are shifted by one uniform offset so the earliest survivor starts at t = 0.traces: 393
main_turns: 28,444
subagent_groups: 1,697
subagent_inner_requests: 39,822
total_model_requests: 68,266
total_input_tokens: 6,891,228,864
total_output_tokens: 58,728,807




python utils/agentic/sample_proxy_traces.py --out '<workdir>/proxy' --sampling top --min-trace-version 7 --max-trace-version 7 --min-requests 20 --require-cli-min 2.1.139 --max-parallel-subagents 10 --exclude-dynamic-workflow-bug
Load in aiperf via:
--public-dataset semianalysis_cc_traces_weka_with_subagents_256k
2 commits