Build from SearchSwarm/SearchSwarm-SFT with scripts/datasets/searchswarm_shared.py

README.md

@@ -82,8 +82,12 @@ One row is one measured request:
 - `subagents[*].messages`: each sub-agent's full conversation, system prompt to final message.
   The final message ends with the report (inside `<report>` tags, after a `<think>` block).
   These are the **store** requests.
-- `main_messages`: the main agent's conversation from its system prompt up to and including
-  the user message that carries the sub-agents' reports. This is the **measured** request.
+- `main_history`: the main agent's conversation up to and including the dispatch turn, which is
+  the request the main agent sent before the reports came back. In production that request was
+  served just before the measured one, so its KV is a **prefix hit** for the measured request;
+  send it first, in every arm, to reproduce that.
+- `main_messages`: `main_history` plus the user message that carries the sub-agents' reports.
+  This is the **measured** request.
   The last message concatenates 2 to 5 reports as
   `A sub-agent dispatched for goal "<goal>" returned the following report:\n\n<report>`.
 - `main_next`: the main agent's following assistant turn in the source trajectory, for
@@ -125,6 +129,7 @@ Exact percentiles are in `stats.json`.
 | `task_index`, `round` | int | source row index; dispatch round within the task (1 = first) |
 | `source`, `question`, `answer` | string | from the source |
 | `n_reports` | int | reports in the reply, equal to `len(subagents)` |
+| `main_history` | list of `{role, content}` | the main agent's previous request: `main_messages` without its last message |
 | `main_messages` | list of `{role, content}` | the measured request |
 | `main_next` | string or null | the main agent's next assistant turn in the source |
 | `subagents` | list of `{goal, brief, report, report_tokens, context_tokens, messages}` | one per report, in reply order |
@@ -135,14 +140,18 @@ Exact percentiles are in `stats.json`.
 ## Intended protocol
 
 Send each `subagents[i].messages` as a store request (the sub-agent's whole context, ending in
-its final message), then send `main_messages` once as the measured request with storing off,
-and compare the main agent's next turn between a blend endpoint and a recompute endpoint.
+its final message) and `main_history` as the main agent's previous request (stored on the
+blend endpoint, plain on the recompute endpoint, so both hold it in the engine's prefix
+cache), then send `main_messages` once as the measured request with storing off, and compare
+the main agent's next turn between a blend endpoint and a recompute endpoint. The measured
+request then has a prefix hit on its own history and non-prefix hits on the reports, which is
+the production shape; the two together are the overall hit rate.
 The main agent's own earlier `<think>` blocks are kept in the record; drop or keep them
 consistently across arms (the served model's chat template usually drops them).
 
 ## Provenance
 
-Built on 2026-11-13 with a deterministic filter script (`searchswarm_shared.py`, part of
+Built on 2026-11-13 (`main_history` added the same day) with a deterministic filter script (`searchswarm_shared.py`, part of
 TensorMesh's CacheBlend evaluation suite) from the source revision current on that date. The source trajectories were generated by the
 SearchSwarm authors' research harness with an unnamed teacher model and kept only when the
 main agent's final answer was correct. Licence follows the source: MIT.

data/train.parquet

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:4a97dbbe23b7a98713217cdc0a300adf601338939f383985993a00f70f683561
-size 371019540
+oid sha256:a6d98ac8d23c99edc56ca2b665d62b7744d886464395fa3658001f146841d03e
+size 375861399