Attempt to improve explain English prompt

.github/workflows/eval-functions.yml

@@ -104,12 +104,12 @@ jobs:
           mkdir -p eval-results
           EVAL_ERRORS=0
 
-          echo "Running explainText evaluation..."
-          genkit eval:flow explainText \
-            --input datasets/explain-chinese.json \
-            --evaluators=custom/chineseTextPresent,custom/validPinyinFormat,custom/outputStructureValid,custom/grammarExplanationQuality \
-            --batchSize 10 \
-            --output eval-results/explain-chinese-results.json || { echo "⚠️ explainText evaluation had errors"; EVAL_ERRORS=$((EVAL_ERRORS+1)); }
+          # echo "Running explainText evaluation..."
+          # genkit eval:flow explainText \
+          #   --input datasets/explain-chinese.json \
+          #   --evaluators=custom/chineseTextPresent,custom/validPinyinFormat,custom/outputStructureValid,custom/grammarExplanationQuality \
+          #   --batchSize 10 \
+          #   --output eval-results/explain-chinese-results.json || { echo "⚠️ explainText evaluation had errors"; EVAL_ERRORS=$((EVAL_ERRORS+1)); }
 
           echo "Running explainEnglish evaluation..."
           genkit eval:flow explainEnglish \
@@ -118,26 +118,26 @@ jobs:
             --batchSize 10 \
             --output eval-results/explain-english-results.json || { echo "⚠️ explainEnglish evaluation had errors"; EVAL_ERRORS=$((EVAL_ERRORS+1)); }
 
-          echo "Running generateChineseSentences evaluation..."
-          genkit eval:flow generateChineseSentences \
-            --input datasets/generate-chinese-sentences.json \
-            --evaluators=custom/chineseTextPresent,custom/validPinyinFormat,custom/outputStructureValid,custom/sentenceGenerationQuality \
-            --batchSize 10 \
-            --output eval-results/generate-sentences-results.json || { echo "⚠️ generateChineseSentences evaluation had errors"; EVAL_ERRORS=$((EVAL_ERRORS+1)); }
-
-          echo "Running analyzeCollocation evaluation..."
-          genkit eval:flow analyzeCollocation \
-            --input datasets/analyze-collocation.json \
-            --evaluators=custom/chineseTextPresent,custom/englishTranslationPresent,custom/outputStructureValid \
-            --batchSize 10 \
-            --output eval-results/collocation-results.json || { echo "⚠️ analyzeCollocation evaluation had errors"; EVAL_ERRORS=$((EVAL_ERRORS+1)); }
-
-          echo "Running explainWordInContext evaluation..."
-          genkit eval:flow explainWordInContext \
-            --input datasets/explain-word-in-context.json \
-            --evaluators=custom/chineseTextPresent,custom/englishTranslationPresent,custom/outputStructureValid \
-            --batchSize 10 \
-            --output eval-results/word-context-results.json || { echo "⚠️ explainWordInContext evaluation had errors"; EVAL_ERRORS=$((EVAL_ERRORS+1)); }
+          # echo "Running generateChineseSentences evaluation..."
+          # genkit eval:flow generateChineseSentences \
+          #   --input datasets/generate-chinese-sentences.json \
+          #   --evaluators=custom/chineseTextPresent,custom/validPinyinFormat,custom/outputStructureValid,custom/sentenceGenerationQuality \
+          #   --batchSize 10 \
+          #   --output eval-results/generate-sentences-results.json || { echo "⚠️ generateChineseSentences evaluation had errors"; EVAL_ERRORS=$((EVAL_ERRORS+1)); }
+
+          # echo "Running analyzeCollocation evaluation..."
+          # genkit eval:flow analyzeCollocation \
+          #   --input datasets/analyze-collocation.json \
+          #   --evaluators=custom/chineseTextPresent,custom/englishTranslationPresent,custom/outputStructureValid \
+          #   --batchSize 10 \
+          #   --output eval-results/collocation-results.json || { echo "⚠️ analyzeCollocation evaluation had errors"; EVAL_ERRORS=$((EVAL_ERRORS+1)); }
+
+          # echo "Running explainWordInContext evaluation..."
+          # genkit eval:flow explainWordInContext \
+          #   --input datasets/explain-word-in-context.json \
+          #   --evaluators=custom/chineseTextPresent,custom/englishTranslationPresent,custom/outputStructureValid \
+          #   --batchSize 10 \
+          #   --output eval-results/word-context-results.json || { echo "⚠️ explainWordInContext evaluation had errors"; EVAL_ERRORS=$((EVAL_ERRORS+1)); }
 
           if [ $EVAL_ERRORS -gt 0 ]; then
             echo "⚠️ $EVAL_ERRORS evaluation(s) had errors - check results for details"

functions/prompts/explain-english.prompt

@@ -7,9 +7,56 @@ output:
   schema: EnglishExplanationSchema
 ---
 {{role "system"}}
-You are a helpful Chinese teacher for speakers of English who want to learn Chinese. You provide clear, concise explanations that help learners understand Chinese.
+You are an expert Chinese language tutor. You have thoroughly studied the Chinese Grammar Wiki and HSK Standard Course textbooks, and you use their terminology and teaching approaches.
+
+Prioritize accuracy over comprehensiveness — only explain what you are confident about.
+Keep your explanations focused and practical for a language learner.
+
+CRITICAL REQUIREMENTS:
+1. Your pinyin, vocabulary breakdown, and grammar explanations must exactly match the Chinese characters in your translation. Do not explain words or characters that are not present in your translation.
+2. When describing grammar structures in your explanation, verify they match the actual translation you provided. Do not claim you used a structure (like 比 or 把) if your translation uses a different one.
+3. For stative sentences (describing states like "the door is open"), remember to use 着 or other appropriate aspect markers.
+
 {{role "user"}}
 Translate the English text input by the user into Chinese, and explain the translation.
 
-Here's the user's English input:
+Here are some examples:
+
+**Example 1:**
+Input: "This is the tallest building I've ever seen"
+Translation: 这是我见过的最高的建筑。
+Pinyin: Zhè shì wǒ jiàn guo de zuì gāo de jiànzhù.
+Explanation: This sentence uses 是 as a simple copula (meaning "is"), NOT the emphatic 是...的 construction. The 的 here is the attributive particle connecting the modifier "我见过" to the noun. The 过 indicates past experience ("have seen").
+
+**Example 2:**
+Input: "I waited for her for two hours"
+Translation: 我等了她两个小时。
+Pinyin: Wǒ děng le tā liǎng gè xiǎoshí.
+Explanation: When the object is a pronoun (她), it comes directly after the verb+了, and the duration (两个小时) goes at the end: Verb + 了 + Pronoun + Duration.
+
+**Example 3:**
+Input: "Rather than watching TV, it's better to read a book"
+Translation: 与其看电视，不如看书。
+Pinyin: Yǔqí kàn diànshì, bùrú kàn shū.
+Explanation: The 与其...不如... structure means "rather than X, it's better to Y." 与其 introduces the less preferred option, 不如 introduces the preferred one. This is different from using 比较好.
+
+**Example 4:**
+Input: "Only she understands"
+Translation: 只有她懂。
+Pinyin: Zhǐyǒu tā dǒng.
+Explanation: 只有 means "only." This sentence does NOT contain 的 — do not explain 的 when it's not present. The structure is simply: 只有 + subject + verb.
+
+**Example 5:**
+Input: "I just got home"
+Translation: 我刚到家。
+Pinyin: Wǒ gāng dào jiā.
+Explanation: 刚 means "just (now)." Note: this translation does NOT use 了 — the 刚 alone conveys the recent completion. Do not explain 了 when it's not in the sentence.
+
+**Example 6:**
+Input: "Even teachers make mistakes"
+Translation: 连老师都会犯错。
+Pinyin: Lián lǎoshī dōu huì fàn cuò.
+Explanation: 连...都... means "even..." for emphasis. This sentence does NOT use 是...的 — it's a simple statement with the 连...都 pattern. Do not mention 是...的 when it's not present.
+
+Now translate and explain this input:
 {{text}}

functions/src/genkit-eval.ts

@@ -38,7 +38,7 @@ const ai = genkit({
             projectId: process.env.GCLOUD_PROJECT || 'hanzigraph',
         }),
     ],
-    model: vertexAI.model('gemini-3-flash-preview'),
+    model: vertexAI.model('gemini-3-pro-preview'),
 });
 
 // Register schemas - MUST be done before loading prompts