e5_summary.json 2.8 KB

1234567891011121314151617181920212223242526272829303132333435363738394041424344454647484950515253545556575859606162636465666768697071727374757677787980818283848586878889909192939495
  1. {
  2. "total_groups": 5,
  3. "rs_cross_stats": {
  4. "mean": 0.1759047640612712,
  5. "std": 0.14271632091758432,
  6. "min": 0.05101916888623692,
  7. "max": 0.4316129517679338
  8. },
  9. "groups_sorted_by_rs_cross": [
  10. {
  11. "name": "group_0_(最短)",
  12. "rs_cross": 0.4316129517679338,
  13. "num_texts": 10,
  14. "avg_length": 329.6,
  15. "r_effs": {
  16. "llama3.2-3b-instruct": 2.226330041885376,
  17. "Mistral-7B-v0.3": 2.4710209369659424,
  18. "llama3-8b": 2.1663646697998047,
  19. "Qwen2.5-7B-Instruct": 3.8829245567321777,
  20. "gemma-2-9b-it": 2.18928599357605
  21. }
  22. },
  23. {
  24. "name": "group_1_(较短)",
  25. "rs_cross": 0.23267617354590583,
  26. "num_texts": 10,
  27. "avg_length": 446.0,
  28. "r_effs": {
  29. "llama3.2-3b-instruct": 2.1320927143096924,
  30. "Mistral-7B-v0.3": 2.282879590988159,
  31. "llama3-8b": 1.977645993232727,
  32. "Qwen2.5-7B-Instruct": 3.2836060523986816,
  33. "gemma-2-9b-it": 2.0314362049102783
  34. }
  35. },
  36. {
  37. "name": "group_2_(中等)",
  38. "rs_cross": 0.08646117594966882,
  39. "num_texts": 10,
  40. "avg_length": 555.8,
  41. "r_effs": {
  42. "llama3.2-3b-instruct": 2.0396993160247803,
  43. "Mistral-7B-v0.3": 2.237123727798462,
  44. "llama3-8b": 1.899621844291687,
  45. "Qwen2.5-7B-Instruct": 2.7096762657165527,
  46. "gemma-2-9b-it": 1.9543957710266113
  47. }
  48. },
  49. {
  50. "name": "group_4_(最长)",
  51. "rs_cross": 0.0777543501566106,
  52. "num_texts": 10,
  53. "avg_length": 859.0,
  54. "r_effs": {
  55. "llama3.2-3b-instruct": 2.1292803287506104,
  56. "Mistral-7B-v0.3": 2.7304728031158447,
  57. "llama3-8b": 2.1836905479431152,
  58. "Qwen2.5-7B-Instruct": 2.7516024112701416,
  59. "gemma-2-9b-it": 2.210625410079956
  60. }
  61. },
  62. {
  63. "name": "group_3_(较长)",
  64. "rs_cross": 0.05101916888623692,
  65. "num_texts": 10,
  66. "avg_length": 684.8,
  67. "r_effs": {
  68. "llama3.2-3b-instruct": 2.1218972206115723,
  69. "Mistral-7B-v0.3": 2.568671703338623,
  70. "llama3-8b": 2.1244308948516846,
  71. "Qwen2.5-7B-Instruct": 2.597262144088745,
  72. "gemma-2-9b-it": 2.120511054992676
  73. }
  74. }
  75. ],
  76. "correlations": {
  77. "rs_cross_vs_text_length": {
  78. "spearman_rho": -0.8999999999999998,
  79. "spearman_pvalue": 0.03738607346849874,
  80. "n_samples": 5,
  81. "interpretation": "正相关表示长文本(复杂)导致更高 RS_cross"
  82. },
  83. "rs_cross_vs_estimated_difficulty": {
  84. "spearman_rho": 0.49999999999999994,
  85. "spearman_pvalue": 0.3910022189557703,
  86. "n_samples": 5,
  87. "interpretation": "正相关表示高难度导致更高 RS_cross"
  88. },
  89. "rs_cross_trend": {
  90. "spearman_rho": -0.8999999999999998,
  91. "spearman_pvalue": 0.03738607346849874,
  92. "interpretation": "正相关表示 RS_cross 随文本长度增加而上升"
  93. }
  94. }
  95. }