Release from the HiveBench Studio (private)

README.md

@@ -33,26 +33,26 @@ this edition adds the encoder measurement taken on the harness's own traffic
 A long conversation is not a long document. Facts that mattered twenty turns
 ago fall out of a rolling context window; the model answers from what
 survives, not from what happened; and the surviving context is mostly noise
-by then. Truncation and recency are cheap and wrong. They throw away
-exactly the old fact the user came back for. Keeping everything is worse;
+by then. Truncation and recency are cheap and wrong: they throw away
+exactly the old fact the user came back for. Keeping everything is worse -
 the prompt grows, the KV cache grows, the cost grows, and the model
-"loses the middle" while the interesting fact *sits somewhere in the middle*.
+"loses the middle" while the interesting fact sits somewhere in the middle.
 
 ## What this does
 
 Splinter Memory keeps a **managed** memory outside the model and hands the
 model a **bounded, relevance-ranked** context every turn:
 
-- A small fleet of **bidirectional encoders** ("drones") scores every stored
-  chunk against the current query; cheap, and far better at similarity than
-  a generative model of the same size.
-- A **membrane** drops what is said twice and notices when the topic
-  changes.
-- A **retention layer** makes forgetting a *design parameter*; an
+- a small fleet of **bidirectional encoders** ("drones") scores every stored
+  chunk against the current query - cheap, and far better at similarity than
+  a generative model of the same size;
+- a **membrane** drops what is said twice and notices when the topic
+  changes;
+- a **retention layer** makes forgetting a *design parameter* - an
   escalating-friction decay, a remembrance pass that rescues facts worth
   keeping, and a "comb" (an SSD tier) that resurrects topics that returned
-  after being evicted.
-- A **focal layer** assembles a bounded context on an adaptive token budget
+  after being evicted;
+- a **focal layer** assembles a bounded context on an adaptive token budget
   and hands it to any OpenAI-compatible backend (llama.cpp, LM Studio, vLLM,
   a hosted API).
 
@@ -60,10 +60,10 @@ The model never sees the raw history. It sees the part that is worth its
 attention, every turn, at a cost that does not grow with the conversation.
 
 **The separation is the point.** The generative model is good at producing
-answers and the wrong tool for selecting context; measured live, *using the
-27B generative model as an encoder-proxy cost tens of seconds per turn on
+answers and the wrong tool for selecting context; measured live, using the
+27B generative model as an encoder proxy cost tens of seconds per turn on
 the hot path, while the 17M-parameter drone did the same job in
-milliseconds.*
+milliseconds.
 
 ## The measured results
 
@@ -130,7 +130,7 @@ generates → the reply is observed back into the store.
 1. **Context curation** - for a fixed token budget, a relevance-ranked
    selection from the full history dominates a contiguous window of the same
    size, and the gap widens with conversation length.
-2. **Managed decay** - forgetting is a design parameter, ***not*** a failure mode;
+2. **Managed decay** - forgetting is a design parameter, not a failure mode;
    escalating friction beats both unbounded growth and uniform forgetting.
 3. **Separation** - bidirectional encoders for comprehension, an
    autoregressive decoder for generation, is the better allocation of the
@@ -173,11 +173,11 @@ The harness gives its encoder three jobs, and only one of them is a ranking:
 The prior encoder comparison measured **rank** only, on the synthetic
 fixtures, and found that six encoders (a 22M MiniLM, a 400MB cross-encoder,
 a 568M retrieval-specialised encoder and two contrastively tuned variants)
-all landed on the same top-K curve; so it concluded the ceiling was
+all landed on the same top-K curve - so it concluded the ceiling was
 *data-structural* and that the remedy was "stop looking for a better
 encoder".
 
-We re-measured all three jobs on the harness's **own** traffic; 81
+We re-measured all three jobs on the harness's **own** traffic - 81
 conversations, 6,975 stored chunks, 4,056 duplicate pairs and 6,533
 topic-change cases, labelled lexically so the label is independent of the
 encoder being measured (`experiments/encoder_suite.py`). The result
@@ -195,7 +195,7 @@ not re-derived.
 | drift at the shipped threshold (0.6) | precision 0.08, **recall 0.96** - the topic-change test fires on almost every turn |
 | drift as a *ranking* signal | **AUC 0.945** - the signal is strong; the threshold is wrong |
 
-The encoder is not the bottleneck. The **operating points are**, and they
+The encoder is not the bottleneck. The **operating points** are - and they
 are per-encoder, so they have to travel with the model. The full sweep
 (model families, sizes, dimensions, CPU cost per chunk) is in
 `docs/ENCODER-NOTES.md` in the harness repository, and the evaluation sets
@@ -214,7 +214,7 @@ are published as a dataset next to this card.
   measurement in §7 is on real traffic; the P1-P12 verdicts mostly are not.
 - **Cross-conversation contamination and hedge-replies** are harness-level
   failure modes that look like architectural failures if you do not
-  isolate conversations per-store and filter refusals *out* of the store.
+  isolate conversations per store and filter refusals out of the store.
 - **The decay curve is an engineering heuristic** with cognitive-science
   inspiration, not a validated model of human forgetting.
 
@@ -250,4 +250,5 @@ MIT. See `LICENSE` and `CITATION.cff` in this repository.
 
 **Community invitation.** The protocol is written to be challenged: every
 prediction has a logic chain, a measurement and a falsification condition.
-If your hardware produces a different verdict, that is a result I am interested in, open a topic/issue with the run bundle.
+If your hardware produces a different verdict, that is a result - open an
+issue with the run bundle.