reasoner_process

TheRoadQaQ · TheRoadQaQ · commit 50531a361075 · 2025-04-07T23:57:14.000+08:00
diff --git a/configs/process/text_process_reasoner_ansfilter.yaml b/configs/process/text_process_reasoner_ansfilter.yaml
@@ -1,6 +1,6 @@
 model_cache_path: '../ckpt' # Path to cache models
 dependencies: [text]
-save_path: "./processed.jsonl"
+save_path: "../dataflow-develop/processed.jsonl"
 
 data:
   text:
@@ -9,21 +9,20 @@ data:
     dataset_split: 'train'  
     name: 'default' 
     revision: null
-    data_path: 'demos/text_process/reasoners/math_5_samples.json'  # Local data path, supports json, jsonl, parquet formats
+    data_path: './demos/text_process/reasoners/math_5_samples.json'  # Local data path, supports json, jsonl, parquet formats
     formatter: "TextFormatter" # Data loader type
     keys: 'answer' # Key name to be processed, for sft data, it can be specified as ['instruction','input','output']
 
 processors:
-  AnswerFormatterFilter:
-    type: "default"
+  AnswerFormatterFilter: {}
   AnswerNgramFilter:
-    min_score: 0.5
+    min_score: 0.1
     max_score: 1.0
     ngrams: 5
   AnswerGroundTruthFilter: 
-    compare_method: exact # exact/math_verify/xverify
+    compare_method: math_verify # exact or math_verify
   AnswerTokenLengthFilter:
-    max_answer_token_length: 1024
+    max_answer_token_length: 512
     tokenizer_dir: '../Qwen2.5-0.5B-Instruct'
 
   
diff --git a/demos/text_process/reasoners/math_5_samples.json b/demos/text_process/reasoners/math_5_samples.json
@@ -7,9 +7,9 @@
     },
     {
         "answer": 
-         "1. Find critical points:\n        → f'(x) = 3x² - 6x\n        → Set derivative to zero: 3x(x-2) = 0 ⇒ x=0, x=2\n    \n        2. Evaluate function at critical points and endpoints:\n        → f(-1) = (-1)^3 - 3(-1)^2 + 4 = -1 -3 +4 = 0.0000\n        → f(0) = 0³ - 3(0)² +4 = 4.0000\n        → f(2) = 8 - 12 +4 = 0.0000\n        → f(3) = 27 - 27 +4 = 4.0000\n    \n        3. Compare values:\n        → Minimum occurs at x=-1 and x=2\n    \n        Verification:\n        → Second derivative test: f''(x) = 6x-6\n        → f''(-1) = -12 < 0 (local max)\n        → f''(2) = 6 > 0 (local min)\n    \n        \\boxed{0}"
+         "Solution:\n        1. Find critical points:\n        → f'(x) = 3x² - 6x\n        → Set derivative to zero: 3x(x-2) = 0 ⇒ x=0, x=2\n    \n        2. Evaluate function at critical points and endpoints:\n        → f(-1) = (-1)^3 - 3(-1)^2 + 4 = -1 -3 +4 = 0.0000\n        → f(0) = 0³ - 3(0)² +4 = 4.0000\n        → f(2) = 8 - 12 +4 = 0.0000\n        → f(3) = 27 - 27 +4 = 4.0000\n    \n        3. Compare values:\n        → Minimum occurs at x=-1 and x=2\n    \n        Verification:\n        → Second derivative test: f''(x) = 6x-6\n        → f''(-1) = -12 < 0 (local max)\n        → f''(2) = 6 > 0 (local min)\n    \n        \\boxed{0.5}"
         ,
-        "ground_truth_answer": "0"
+        "ground_truth_answer": "1/2"
     },
     {
         "answer": 
@@ -30,7 +30,7 @@
     },
     {
         "answer": 
-         "Solution:\n        1. Find critical points:\n        → f'(x) = 3x² - 6x\n  1. Find critical points:\n        → f'(x) = 3x² - 6x\n1. Find critical points:\n        → f'(x) = 3x² - 6x\n1. Find critical points:\n        → f'(x) = 3x² - 6x\n1. Find critical points:\n        → f'(x) = 3x² - 6x\n1. Find critical points:\n        → f'(x) = 3x² - 6x\n1. Find critical points:\n        → f'(x) = 3x² - 6x\n1. Find critical points:\n        → f'(x) = 3x² - 6x\n1. Find critical points:\n        → f'(x) = 3x² - 6x\n1. Find critical points:\n        → f'(x) = 3x² - 6x\n1. Find critical points:\n        → f'(x) = 3x² - 6x\n1. Find critical points:\n        → f'(x) = 3x² - 6x\n1. Find critical points:\n        → f'(x) = 3x² - 6x\n1. Find critical points:\n        → f'(x) = 3x² - 6x\n1. Find critical points:\n        → f'(x) = 3x² - 6x\n1. Find critical points:\n        → f'(x) = 3x² - 6x\n1. Find critical points:\n        → f'(x) = 3x² - 6x\n1. Find critical points:\n        → f'(x) = 3x² - 6x\n1. Find critical points:\n        → f'(x) = 3x² - 6x\n1. Find critical points:\n        → f'(x) = 3x² - 6x\n1. Find critical points:\n        → f'(x) = 3x² - 6x\n1. Find critical points:\n        → f'(x) = 3x² - 6x\n1. Find critical points:\n        → f'(x) = 3x² - 6x\n1. Find critical points:\n        → f'(x) = 3x² - 6x\n1. Find critical points:\n        → f'(x) = 3x² - 6x\n1. Find critical points:\n        → f'(x) = 3x² - 6x\n1. Find critical points:\n        → f'(x) = 3x² - 6x\n1. Find critical points:\n        → f'(x) = 3x² - 6x\n1. Find critical points:\n        → f'(x) = 3x² - 6x\n      → Set derivative to zero: 3x(x-2) = 0 ⇒ x=0, x=2\n    \n        2. Evaluate function at critical points and endpoints:\n        → f(-1) = (-1)^3 - 3(-1)^2 + 4 = -1 -3 +4 = 0.0000\n        → f(0) = 0³ - 3(0)² +4 = 4.0000\n        → f(2) = 8 - 12 +4 = 0.0000\n        → f(3) = 27 - 27 +4 = 4.0000\n    \n        3. Compare values:\n        → Minimum occurs at x=-1 and x=2\n    \n        Verification:\n        → Second derivative test: f''(x) = 6x-6\n        → f''(-1) = -12 < 0 (local max)\n        → f''(2) = 6 > 0 (local min)\n    \n        \\boxed{0}"
+         "Solution:\n        1. Find critical points:\n        → f'(x) = 3x² - 6x\n  1. Find critical points:\n        → f'(x) = 3x² - 6x\n1. Find critical points:\n        → f'(x) = 3x² - 6x\n1. Find critical points:\n        → f'(x) = 3x² - 6x\n1. Find critical points:\n        → f'(x) = 3x² - 6x\n1. Find critical points:\n        → f'(x) = 3x² - 6x\n1. Find critical points:\n        → f'(x) = 3x² - 6x\n1. Find critical points:\n        → f'(x) = 3x² - 6x\n1. Find critical points:\n        → f'(x) = 3x² - 6x\n1. Find critical points:\n        → f'(x) = 3x² - 6x\n1. Find critical points:\n        → f'(x) = 3x² - 6x\n1. Find critical points:\n        → f'(x) = 3x² - 6x\n1. Find critical points:\n        → f'(x) = 3x² - 6x\n1. Find critical points:\n        → f'(x) = 3x² - 6x\n1. Find critical points:\n        → f'(x) = 3x² - 6x\n1. Find critical points:\n        → f'(x) = 3x² - 6x\n1. Find critical points:\n        → f'(x) = 3x² - 6x\n1. Find critical points:\n        → f'(x) = 3x² - 6x\n1. Find critical points:\n        → f'(x) = 3x² - 6x\n1. Find critical points:\n        → f'(x) = 3x² - 6x\n1. Find critical points:\n        → f'(x) = 3x² - 6x\n1. Find critical points:\n        → f'(x) = 3x² - 6x\n1. Find critical points:\n        → f'(x) = 3x² - 6x\n1. Find critical points:\n        → f'(x) = 3x² - 6x\n1. Find critical points:\n        → f'(x) = 3x² - 6x\n1. Find critical points:\n        → f'(x) = 3x² - 6x\n1. Find critical points:\n        → f'(x) = 3x² - 6x\n1. Find critical points:\n        → f'(x) = 3x² - 6x\n1. Find critical points:\n        → f'(x) = 3x² - 6x\n      → Set derivative to zero: 3x(x-2) = 0 ⇒ x=0, x=2  → Set derivative to zero: 3x(x-2) = 0 ⇒ x=0, x=2 → Set derivative to zero: 3x(x-2) = 0 ⇒ x=0, x=2 → Set derivative to zero: 3x(x-2) = 0 ⇒ x=0, x=2 → Set derivative to zero: 3x(x-2) = 0 ⇒ x=0, x=2 → Set derivative to zero: 3x(x-2) = 0 ⇒ x=0, x=2 → Set derivative to zero: 3x(x-2) = 0 ⇒ x=0, x=2 → Set derivative to zero: 3x(x-2) = 0 ⇒ x=0, x=2 → Set derivative to zero: 3x(x-2) = 0 ⇒ x=0, x=2 → Set derivative to zero: 3x(x-2) = 0 ⇒ x=0, x=2 → Set derivative to zero: 3x(x-2) = 0 ⇒ x=0, x=2\n    \n        2. Evaluate function at critical points and endpoints:\n        → f(-1) = (-1)^3 - 3(-1)^2 + 4 = -1 -3 +4 = 0.0000\n        → f(0) = 0³ - 3(0)² +4 = 4.0000\n        → f(2) = 8 - 12 +4 = 0.0000\n        → f(3) = 27 - 27 +4 = 4.0000\n    \n        3. Compare values:\n        → Minimum occurs at x=-1 and x=2\n    \n        Verification:\n        → Second derivative test: f''(x) = 6x-6\n        → f''(-1) = -12 < 0 (local max)\n        → f''(2) = 6 > 0 (local min)\n    \n        \\boxed{0}"
         ,
         "ground_truth_answer": "0"
     }

Original file line number	Diff line number	Diff line change
`@@ -7,9 +7,9 @@`
`7`	`7`	`},`
`8`	`8`	`{`
`9`	`9`	`"answer":`
`10`		- "1. Find critical points:\n → f'(x) = 3x² - 6x\n → Set derivative to zero: 3x(x-2) = 0 ⇒ x=0, x=2\n \n 2. Evaluate function at critical points and endpoints:\n → f(-1) = (-1)^3 - 3(-1)^2 + 4 = -1 -3 +4 = 0.0000\n → f(0) = 0³ - 3(0)² +4 = 4.0000\n → f(2) = 8 - 12 +4 = 0.0000\n → f(3) = 27 - 27 +4 = 4.0000\n \n 3. Compare values:\n → Minimum occurs at x=-1 and x=2\n \n Verification:\n → Second derivative test: f''(x) = 6x-6\n → f''(-1) = -12 < 0 (local max)\n → f''(2) = 6 > 0 (local min)\n \n \\boxed{0}"
	`10`	+ "Solution:\n 1. Find critical points:\n → f'(x) = 3x² - 6x\n → Set derivative to zero: 3x(x-2) = 0 ⇒ x=0, x=2\n \n 2. Evaluate function at critical points and endpoints:\n → f(-1) = (-1)^3 - 3(-1)^2 + 4 = -1 -3 +4 = 0.0000\n → f(0) = 0³ - 3(0)² +4 = 4.0000\n → f(2) = 8 - 12 +4 = 0.0000\n → f(3) = 27 - 27 +4 = 4.0000\n \n 3. Compare values:\n → Minimum occurs at x=-1 and x=2\n \n Verification:\n → Second derivative test: f''(x) = 6x-6\n → f''(-1) = -12 < 0 (local max)\n → f''(2) = 6 > 0 (local min)\n \n \\boxed{0.5}"
`11`	`11`	`,`
`12`		`- "ground_truth_answer": "0"`
	`12`	`+ "ground_truth_answer": "1/2"`
`13`	`13`	`},`
`14`	`14`	`{`
`15`	`15`	`"answer":`
`@@ -30,7 +30,7 @@`
`30`	`30`	`},`
`31`	`31`	`{`
`32`	`32`	`"answer":`
`33`		- "Solution:\n 1. Find critical points:\n → f'(x) = 3x² - 6x\n 1. Find critical points:\n → f'(x) = 3x² - 6x\n1. Find critical points:\n → f'(x) = 3x² - 6x\n1. Find critical points:\n → f'(x) = 3x² - 6x\n1. Find critical points:\n → f'(x) = 3x² - 6x\n1. Find critical points:\n → f'(x) = 3x² - 6x\n1. Find critical points:\n → f'(x) = 3x² - 6x\n1. Find critical points:\n → f'(x) = 3x² - 6x\n1. Find critical points:\n → f'(x) = 3x² - 6x\n1. Find critical points:\n → f'(x) = 3x² - 6x\n1. Find critical points:\n → f'(x) = 3x² - 6x\n1. Find critical points:\n → f'(x) = 3x² - 6x\n1. Find critical points:\n → f'(x) = 3x² - 6x\n1. Find critical points:\n → f'(x) = 3x² - 6x\n1. Find critical points:\n → f'(x) = 3x² - 6x\n1. Find critical points:\n → f'(x) = 3x² - 6x\n1. Find critical points:\n → f'(x) = 3x² - 6x\n1. Find critical points:\n → f'(x) = 3x² - 6x\n1. Find critical points:\n → f'(x) = 3x² - 6x\n1. Find critical points:\n → f'(x) = 3x² - 6x\n1. Find critical points:\n → f'(x) = 3x² - 6x\n1. Find critical points:\n → f'(x) = 3x² - 6x\n1. Find critical points:\n → f'(x) = 3x² - 6x\n1. Find critical points:\n → f'(x) = 3x² - 6x\n1. Find critical points:\n → f'(x) = 3x² - 6x\n1. Find critical points:\n → f'(x) = 3x² - 6x\n1. Find critical points:\n → f'(x) = 3x² - 6x\n1. Find critical points:\n → f'(x) = 3x² - 6x\n1. Find critical points:\n → f'(x) = 3x² - 6x\n → Set derivative to zero: 3x(x-2) = 0 ⇒ x=0, x=2\n \n 2. Evaluate function at critical points and endpoints:\n → f(-1) = (-1)^3 - 3(-1)^2 + 4 = -1 -3 +4 = 0.0000\n → f(0) = 0³ - 3(0)² +4 = 4.0000\n → f(2) = 8 - 12 +4 = 0.0000\n → f(3) = 27 - 27 +4 = 4.0000\n \n 3. Compare values:\n → Minimum occurs at x=-1 and x=2\n \n Verification:\n → Second derivative test: f''(x) = 6x-6\n → f''(-1) = -12 < 0 (local max)\n → f''(2) = 6 > 0 (local min)\n \n \\boxed{0}"
	`33`	+ "Solution:\n 1. Find critical points:\n → f'(x) = 3x² - 6x\n 1. Find critical points:\n → f'(x) = 3x² - 6x\n1. Find critical points:\n → f'(x) = 3x² - 6x\n1. Find critical points:\n → f'(x) = 3x² - 6x\n1. Find critical points:\n → f'(x) = 3x² - 6x\n1. Find critical points:\n → f'(x) = 3x² - 6x\n1. Find critical points:\n → f'(x) = 3x² - 6x\n1. Find critical points:\n → f'(x) = 3x² - 6x\n1. Find critical points:\n → f'(x) = 3x² - 6x\n1. Find critical points:\n → f'(x) = 3x² - 6x\n1. Find critical points:\n → f'(x) = 3x² - 6x\n1. Find critical points:\n → f'(x) = 3x² - 6x\n1. Find critical points:\n → f'(x) = 3x² - 6x\n1. Find critical points:\n → f'(x) = 3x² - 6x\n1. Find critical points:\n → f'(x) = 3x² - 6x\n1. Find critical points:\n → f'(x) = 3x² - 6x\n1. Find critical points:\n → f'(x) = 3x² - 6x\n1. Find critical points:\n → f'(x) = 3x² - 6x\n1. Find critical points:\n → f'(x) = 3x² - 6x\n1. Find critical points:\n → f'(x) = 3x² - 6x\n1. Find critical points:\n → f'(x) = 3x² - 6x\n1. Find critical points:\n → f'(x) = 3x² - 6x\n1. Find critical points:\n → f'(x) = 3x² - 6x\n1. Find critical points:\n → f'(x) = 3x² - 6x\n1. Find critical points:\n → f'(x) = 3x² - 6x\n1. Find critical points:\n → f'(x) = 3x² - 6x\n1. Find critical points:\n → f'(x) = 3x² - 6x\n1. Find critical points:\n → f'(x) = 3x² - 6x\n1. Find critical points:\n → f'(x) = 3x² - 6x\n → Set derivative to zero: 3x(x-2) = 0 ⇒ x=0, x=2 → Set derivative to zero: 3x(x-2) = 0 ⇒ x=0, x=2 → Set derivative to zero: 3x(x-2) = 0 ⇒ x=0, x=2 → Set derivative to zero: 3x(x-2) = 0 ⇒ x=0, x=2 → Set derivative to zero: 3x(x-2) = 0 ⇒ x=0, x=2 → Set derivative to zero: 3x(x-2) = 0 ⇒ x=0, x=2 → Set derivative to zero: 3x(x-2) = 0 ⇒ x=0, x=2 → Set derivative to zero: 3x(x-2) = 0 ⇒ x=0, x=2 → Set derivative to zero: 3x(x-2) = 0 ⇒ x=0, x=2 → Set derivative to zero: 3x(x-2) = 0 ⇒ x=0, x=2 → Set derivative to zero: 3x(x-2) = 0 ⇒ x=0, x=2\n \n 2. Evaluate function at critical points and endpoints:\n → f(-1) = (-1)^3 - 3(-1)^2 + 4 = -1 -3 +4 = 0.0000\n → f(0) = 0³ - 3(0)² +4 = 4.0000\n → f(2) = 8 - 12 +4 = 0.0000\n → f(3) = 27 - 27 +4 = 4.0000\n \n 3. Compare values:\n → Minimum occurs at x=-1 and x=2\n \n Verification:\n → Second derivative test: f''(x) = 6x-6\n → f''(-1) = -12 < 0 (local max)\n → f''(2) = 6 > 0 (local min)\n \n \\boxed{0}"
`34`	`34`	`,`
`35`	`35`	`"ground_truth_answer": "0"`
`36`	`36`	`}`