Update Automated MNLP evaluation report (2026-05-20)

EVAL_REPORT.md

@@ -2,7 +2,7 @@
 
 - **Model repo:** [`cs-552-2026-kth/general_knowledge_model`](https://huggingface.co/cs-552-2026-kth/general_knowledge_model)
 - **Owner(s):** group **kth**
-- **Generated at:** 2026-05-19T03:22:52+00:00 (UTC)
+- **Generated at:** 2026-05-20T05:49:46+00:00 (UTC)
 - **Pipeline:** [mnlp-project-ci](https://github.com/eric11eca/mnlp-project-ci)
 
 _This PR is opened automatically by the course CI. It is **non-blocking** — you do not need to merge it. The next nightly run will refresh this file._
@@ -12,7 +12,7 @@ _This PR is opened automatically by the course CI. It is **non-blocking** — yo
 | Benchmark | Accuracy | Status |
 |---|---:|---|
 | Math | — | not run |
-| Knowledge | 0.3700 | ok |
+| Knowledge | 0.3500 | ok |
 | Multilingual | — | not run |
 | Safety | — | not run |
 
@@ -30,16 +30,16 @@ _Prompts are intentionally omitted to avoid revealing benchmark contents. For mu
 - **completion**:
 
     ```text
-    Answer: \boxed{C}
+    \boxed{C}
     ```
 
 **Incorrect** (1 shown)
 
-- **reference**: `C`
+- **reference**: `B`
 - **overall** (0/1 completions correct)
-- **extracted** (✗): `D`
+- **extracted** (✗): `A`
 - **completion**:
 
     ```text
-    \boxed{D}
+    \boxed{A}
     ```