Update README.md
Browse files
README.md
CHANGED
|
@@ -9,3 +9,83 @@ This version of Solar-10.7B was lasered and perplexity was calculated against gs
|
|
| 9 |
+ New baseline perplexity: 12.554274559020996
|
| 10 |
|
| 11 |
The laser process decreased perplexity by 2.41%
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 9 |
+ New baseline perplexity: 12.554274559020996
|
| 10 |
|
| 11 |
The laser process decreased perplexity by 2.41%
|
| 12 |
+
| Model |AGIEval|GPT4All|TruthfulQA|Bigbench|Average|
|
| 13 |
+
|-----------------------------------------------------------------------------------------------------|------:|------:|---------:|-------:|------:|
|
| 14 |
+
|[SOLAR-10.7B-Instruct-v1.0-laser](https://huggingface.co/macadeliccc/SOLAR-10.7B-Instruct-v1.0-laser)| 46.9| 74.99| 70.64| 43.74| 59.07|
|
| 15 |
+
|
| 16 |
+
### AGIEval
|
| 17 |
+
| Task |Version| Metric |Value| |Stderr|
|
| 18 |
+
|------------------------------|------:|--------|----:|---|-----:|
|
| 19 |
+
|agieval_aqua_rat | 0|acc |29.53|± | 2.87|
|
| 20 |
+
| | |acc_norm|28.35|± | 2.83|
|
| 21 |
+
|agieval_logiqa_en | 0|acc |39.78|± | 1.92|
|
| 22 |
+
| | |acc_norm|40.55|± | 1.93|
|
| 23 |
+
|agieval_lsat_ar | 0|acc |23.04|± | 2.78|
|
| 24 |
+
| | |acc_norm|21.30|± | 2.71|
|
| 25 |
+
|agieval_lsat_lr | 0|acc |51.18|± | 2.22|
|
| 26 |
+
| | |acc_norm|51.76|± | 2.21|
|
| 27 |
+
|agieval_lsat_rc | 0|acc |66.54|± | 2.88|
|
| 28 |
+
| | |acc_norm|66.91|± | 2.87|
|
| 29 |
+
|agieval_sat_en | 0|acc |78.16|± | 2.89|
|
| 30 |
+
| | |acc_norm|78.16|± | 2.89|
|
| 31 |
+
|agieval_sat_en_without_passage| 0|acc |50.97|± | 3.49|
|
| 32 |
+
| | |acc_norm|50.00|± | 3.49|
|
| 33 |
+
|agieval_sat_math | 0|acc |42.73|± | 3.34|
|
| 34 |
+
| | |acc_norm|38.18|± | 3.28|
|
| 35 |
+
|
| 36 |
+
Average: 46.9%
|
| 37 |
+
|
| 38 |
+
### GPT4All
|
| 39 |
+
| Task |Version| Metric |Value| |Stderr|
|
| 40 |
+
|-------------|------:|--------|----:|---|-----:|
|
| 41 |
+
|arc_challenge| 0|acc |60.84|± | 1.43|
|
| 42 |
+
| | |acc_norm|63.99|± | 1.40|
|
| 43 |
+
|arc_easy | 0|acc |83.59|± | 0.76|
|
| 44 |
+
| | |acc_norm|81.44|± | 0.80|
|
| 45 |
+
|boolq | 1|acc |87.58|± | 0.58|
|
| 46 |
+
|hellaswag | 0|acc |68.11|± | 0.47|
|
| 47 |
+
| | |acc_norm|85.77|± | 0.35|
|
| 48 |
+
|openbookqa | 0|acc |35.40|± | 2.14|
|
| 49 |
+
| | |acc_norm|48.40|± | 2.24|
|
| 50 |
+
|piqa | 0|acc |80.58|± | 0.92|
|
| 51 |
+
| | |acc_norm|80.74|± | 0.92|
|
| 52 |
+
|winogrande | 0|acc |77.03|± | 1.18|
|
| 53 |
+
|
| 54 |
+
Average: 74.99%
|
| 55 |
+
|
| 56 |
+
### TruthfulQA
|
| 57 |
+
| Task |Version|Metric|Value| |Stderr|
|
| 58 |
+
|-------------|------:|------|----:|---|-----:|
|
| 59 |
+
|truthfulqa_mc| 1|mc1 |55.45|± | 1.74|
|
| 60 |
+
| | |mc2 |70.64|± | 1.49|
|
| 61 |
+
|
| 62 |
+
Average: 70.64%
|
| 63 |
+
|
| 64 |
+
### Bigbench
|
| 65 |
+
| Task |Version| Metric |Value| |Stderr|
|
| 66 |
+
|------------------------------------------------|------:|---------------------|----:|---|-----:|
|
| 67 |
+
|bigbench_causal_judgement | 0|multiple_choice_grade|57.37|± | 3.60|
|
| 68 |
+
|bigbench_date_understanding | 0|multiple_choice_grade|62.87|± | 2.52|
|
| 69 |
+
|bigbench_disambiguation_qa | 0|multiple_choice_grade|35.66|± | 2.99|
|
| 70 |
+
|bigbench_geometric_shapes | 0|multiple_choice_grade|33.15|± | 2.49|
|
| 71 |
+
| | |exact_str_match | 0.00|± | 0.00|
|
| 72 |
+
|bigbench_logical_deduction_five_objects | 0|multiple_choice_grade|26.20|± | 1.97|
|
| 73 |
+
|bigbench_logical_deduction_seven_objects | 0|multiple_choice_grade|19.71|± | 1.50|
|
| 74 |
+
|bigbench_logical_deduction_three_objects | 0|multiple_choice_grade|45.00|± | 2.88|
|
| 75 |
+
|bigbench_movie_recommendation | 0|multiple_choice_grade|39.00|± | 2.18|
|
| 76 |
+
|bigbench_navigate | 0|multiple_choice_grade|51.20|± | 1.58|
|
| 77 |
+
|bigbench_reasoning_about_colored_objects | 0|multiple_choice_grade|53.90|± | 1.11|
|
| 78 |
+
|bigbench_ruin_names | 0|multiple_choice_grade|40.18|± | 2.32|
|
| 79 |
+
|bigbench_salient_translation_error_detection | 0|multiple_choice_grade|39.98|± | 1.55|
|
| 80 |
+
|bigbench_snarks | 0|multiple_choice_grade|63.54|± | 3.59|
|
| 81 |
+
|bigbench_sports_understanding | 0|multiple_choice_grade|68.36|± | 1.48|
|
| 82 |
+
|bigbench_temporal_sequences | 0|multiple_choice_grade|65.20|± | 1.51|
|
| 83 |
+
|bigbench_tracking_shuffled_objects_five_objects | 0|multiple_choice_grade|22.48|± | 1.18|
|
| 84 |
+
|bigbench_tracking_shuffled_objects_seven_objects| 0|multiple_choice_grade|18.46|± | 0.93|
|
| 85 |
+
|bigbench_tracking_shuffled_objects_three_objects| 0|multiple_choice_grade|45.00|± | 2.88|
|
| 86 |
+
|
| 87 |
+
Average: 43.74%
|
| 88 |
+
|
| 89 |
+
Average score: 59.07%
|
| 90 |
+
|
| 91 |
+
Elapsed time: 02:33:24
|