Update README

README.md

@@ -37,7 +37,7 @@ tags:
 - **84.6% macro-average balanced accuracy** across the same 9 benchmarks
 - Out-of-domain: **60.2% balanced accuracy on Coverbench**, **77.0% on LLM-AggreFact**
 - Strong on long-form evidence: 87% on Ex-FEVER, 92% on FEVEROUS, 76% on HoVer
-- Reasoning is **fully transparent** — the model emits its sub-claim checklist, every question it asked, every quote from evidence, and a final label
+- Reasoning is **fully transparent**: the model emits its sub-claim checklist, every question it asked, every quote from evidence, and a final label
 
 ## Model Overview
 
@@ -60,28 +60,28 @@ DecomposeRL trains the policy to follow a **decompose-question-answer-verify** l
 3. **Sufficiency check** (`<think>`): track which sub-claims are resolved; loop until every sub-claim is addressed.
 4. **Final verdict** (`<verification>`): `Supported` or `Refuted`.
 
-### Reward Stack — seven complementary signals
+### Reward Stack: seven complementary signals
 
 GRPO is supervised with a sum of seven rewards, grouped into three families:
 
 **Programmatic anchors** (no judge call)
 
-1. **Format** — ensures the trace is parseable; a gating prerequisite without which no other reward can be computed.
-2. **Question count** — discourages collapsing the decomposition into one mega-question or padding it with filler.
-3. **Diversity** — penalizes redundant questions so the policy covers distinct sub-claims instead of rewording the same one.
+1. **Format**: ensures the trace is parseable; a gating prerequisite without which no other reward can be computed.
+2. **Question count**: discourages collapsing the decomposition into one mega-question or padding it with filler.
+3. **Diversity**: penalizes redundant questions so the policy covers distinct sub-claims instead of rewording the same one.
 
 **Set-level signals**
 
-4. **Coverage** — checks whether the verdict can be recovered from the answers alone; tests if the decomposition is *collectively sufficient*.
-5. **Verification** — direct outcome anchor: did the final label match the gold label?
+4. **Coverage**: checks whether the verdict can be recovered from the answers alone; tests if the decomposition is *collectively sufficient*.
+5. **Verification**: direct outcome anchor; did the final label match the gold label?
 
 **Leave-one-out and per-question composites**
 
-6. **Necessity (leave-one-out)** — the only signal that can push the policy to *remove* misleading questions; a question is necessary iff its removal would change the verdict.
-7. **Joint multiplicative quality** — composes three per-question sub-signals so a question must clear *all* of them simultaneously rather than scoring partial credit:
-   - **(7a) Answerability** — is the question answerable from the evidence?
-   - **(7b) Atomicity** — is it a single-focus, verifiable question grounded in the claim?
-   - **(7c) Answer correctness** — is the answer faithful to the document (no contradictions, no extrinsic info)?
+6. **Necessity (leave-one-out)**: the only signal that can push the policy to *remove* misleading questions; a question is necessary iff its removal would change the verdict.
+7. **Joint multiplicative quality**: composes three per-question sub-signals so a question must clear *all* of them simultaneously rather than scoring partial credit:
+   - **(7a) Answerability**: is the question answerable from the evidence?
+   - **(7b) Atomicity**: is it a single-focus, verifiable question grounded in the claim?
+   - **(7c) Answer correctness**: is the answer faithful to the document (no contradictions, no extrinsic info)?
 
 ## Quickstart
 
@@ -213,7 +213,7 @@ the specific companies. Next, verify the companies and the layoffs.
 💬  A2: The evidence document states that Nicholson worked as a consultant for
        companies that laid off nearly 1,900 people since 2015, shutting down
        plants in Wisconsin and other states. But it also says Baldwin cites no
-       evidence that Nicholson's work caused the layoffs and shutdowns — only
+       evidence that Nicholson's work caused the layoffs and shutdowns, only
        some element of truth, our definition of Mostly False.
 
 ──────────────────────────────────────────────────────────────────────────────
@@ -238,7 +238,7 @@ The `--max-model-len` matches the training-time `max_seq_length=16016` (with `ma
 
 ## Performance
 
-### In-domain — balanced accuracy (%) on 9 claim-verification benchmarks
+### In-domain: balanced accuracy (%) on 9 claim-verification benchmarks
 
 Compared against every same-size (Qwen-7B) baseline plus MiniCheck. *Micro* is pooled balanced accuracy across all in-domain samples; *Macro* is the uniform mean across the 9 datasets. **Bold** marks the column winner; *italic* marks the second-best.
 
@@ -269,7 +269,7 @@ Compared against every same-size (Qwen-7B) baseline plus MiniCheck. *Micro* is p
 - **In-scope**: verifying factual claims against a *provided* evidence document (open-book fact verification, retrieval-augmented fact-checking pipelines).
 - **Out-of-scope**: closed-book fact-checking, claim verification against the model's parametric knowledge, real-time news verification without supplied evidence.
 
-The model is trained to say *"I don't know"* when the evidence document is silent — please respect that signal in downstream systems instead of forcing a label.
+The model is trained to say *"I don't know"* when the evidence document is silent; please respect that signal in downstream systems instead of forcing a label.
 
 ## Citation