| { |
| "args": { |
| "repo": "cfahlgren1/tinyfix-realword", |
| "data_repo": "cfahlgren1/tinyfix-train-data", |
| "data_dir": null, |
| "run_name": "letters-v2", |
| "fineweb_tokens": 70000000, |
| "smoltalk_tokens": 50000000, |
| "oasst_tokens": 1000000000000, |
| "vocabulary": 20000, |
| "embedding": 64, |
| "hidden": 256, |
| "steps": 40000, |
| "batch": 4096, |
| "learning_rate": 0.002, |
| "misspellings_per_word": 16, |
| "min_count": 5, |
| "workers": 48, |
| "tuning_pairs": 6000, |
| "output": "/tmp/tinyfix-letters", |
| "no_upload": false |
| }, |
| "parameters": 1960537, |
| "train_minutes": 37.7, |
| "validation": { |
| "accuracy": 0.9536, |
| "corrupted_accuracy": 0.928 |
| }, |
| "real_typos": { |
| "torch": { |
| "careful": { |
| "thresholds": { |
| "known_threshold": 7.0, |
| "unknown_threshold": 7.0, |
| "dev_gain": 2572 |
| }, |
| "dev": { |
| "misspelling": { |
| "right": 0.6022, |
| "wrong": 0.024, |
| "total": 4243 |
| }, |
| "real word": { |
| "right": 0.0803, |
| "wrong": 0.0017, |
| "total": 1757 |
| }, |
| "all right": 0.4493, |
| "all wrong": 0.0175, |
| "email changed per 1000": 1.133, |
| "github changed per 1000": 1.599 |
| }, |
| "test": { |
| "misspelling": { |
| "right": 0.6038, |
| "wrong": 0.0255, |
| "total": 2938 |
| }, |
| "real word": { |
| "right": 0.0697, |
| "wrong": 0.0075, |
| "total": 1062 |
| }, |
| "all right": 0.462, |
| "all wrong": 0.0208, |
| "email changed per 1000": 0.766, |
| "github changed per 1000": 1.0 |
| } |
| }, |
| "eager": { |
| "thresholds": { |
| "known_threshold": 5.0, |
| "unknown_threshold": 5.5, |
| "dev_gain": 3023 |
| }, |
| "dev": { |
| "misspelling": { |
| "right": 0.6872, |
| "wrong": 0.0358, |
| "total": 4243 |
| }, |
| "real word": { |
| "right": 0.1656, |
| "wrong": 0.0074, |
| "total": 1757 |
| }, |
| "all right": 0.5345, |
| "all wrong": 0.0275, |
| "email changed per 1000": 1.866, |
| "github changed per 1000": 2.398 |
| }, |
| "test": { |
| "misspelling": { |
| "right": 0.6923, |
| "wrong": 0.0412, |
| "total": 2938 |
| }, |
| "real word": { |
| "right": 0.1685, |
| "wrong": 0.0113, |
| "total": 1062 |
| }, |
| "all right": 0.5533, |
| "all wrong": 0.0333, |
| "email changed per 1000": 1.066, |
| "github changed per 1000": 2.066 |
| } |
| } |
| }, |
| "matmul-gather": { |
| "careful": { |
| "thresholds": { |
| "known_threshold": 10.5, |
| "unknown_threshold": 7.5, |
| "dev_gain": 2656 |
| }, |
| "dev": { |
| "misspelling": { |
| "right": 0.6462, |
| "wrong": 0.0356, |
| "total": 4243 |
| }, |
| "real word": { |
| "right": 0.0507, |
| "wrong": 0.0028, |
| "total": 1757 |
| }, |
| "all right": 0.4718, |
| "all wrong": 0.026, |
| "email changed per 1000": 1.133, |
| "github changed per 1000": 2.265 |
| }, |
| "test": { |
| "misspelling": { |
| "right": 0.6317, |
| "wrong": 0.0453, |
| "total": 2938 |
| }, |
| "real word": { |
| "right": 0.0452, |
| "wrong": 0.0009, |
| "total": 1062 |
| }, |
| "all right": 0.476, |
| "all wrong": 0.0335, |
| "email changed per 1000": 0.833, |
| "github changed per 1000": 1.4 |
| } |
| }, |
| "eager": { |
| "thresholds": { |
| "known_threshold": 11.5, |
| "unknown_threshold": 5.5, |
| "dev_gain": 2845 |
| }, |
| "dev": { |
| "misspelling": { |
| "right": 0.7165, |
| "wrong": 0.0556, |
| "total": 4243 |
| }, |
| "real word": { |
| "right": 0.0364, |
| "wrong": 0.0023, |
| "total": 1757 |
| }, |
| "all right": 0.5173, |
| "all wrong": 0.04, |
| "email changed per 1000": 1.866, |
| "github changed per 1000": 3.464 |
| }, |
| "test": { |
| "misspelling": { |
| "right": 0.7243, |
| "wrong": 0.063, |
| "total": 2938 |
| }, |
| "real word": { |
| "right": 0.0282, |
| "wrong": 0.0009, |
| "total": 1062 |
| }, |
| "all right": 0.5395, |
| "all wrong": 0.0465, |
| "email changed per 1000": 1.033, |
| "github changed per 1000": 2.266 |
| } |
| } |
| }, |
| "matmul": { |
| "careful": { |
| "thresholds": { |
| "known_threshold": 10.5, |
| "unknown_threshold": 7.5, |
| "dev_gain": 2608 |
| }, |
| "dev": { |
| "misspelling": { |
| "right": 0.6356, |
| "wrong": 0.0368, |
| "total": 4243 |
| }, |
| "real word": { |
| "right": 0.0518, |
| "wrong": 0.0028, |
| "total": 1757 |
| }, |
| "all right": 0.4647, |
| "all wrong": 0.0268, |
| "email changed per 1000": 1.166, |
| "github changed per 1000": 2.199 |
| }, |
| "test": { |
| "misspelling": { |
| "right": 0.6341, |
| "wrong": 0.0439, |
| "total": 2938 |
| }, |
| "real word": { |
| "right": 0.0452, |
| "wrong": 0.0028, |
| "total": 1062 |
| }, |
| "all right": 0.4778, |
| "all wrong": 0.033, |
| "email changed per 1000": 0.866, |
| "github changed per 1000": 1.4 |
| } |
| }, |
| "eager": { |
| "thresholds": { |
| "known_threshold": 8.5, |
| "unknown_threshold": 6.5, |
| "dev_gain": 2822 |
| }, |
| "dev": { |
| "misspelling": { |
| "right": 0.6778, |
| "wrong": 0.0481, |
| "total": 4243 |
| }, |
| "real word": { |
| "right": 0.1019, |
| "wrong": 0.0057, |
| "total": 1757 |
| }, |
| "all right": 0.5092, |
| "all wrong": 0.0357, |
| "email changed per 1000": 1.866, |
| "github changed per 1000": 2.865 |
| }, |
| "test": { |
| "misspelling": { |
| "right": 0.6872, |
| "wrong": 0.0511, |
| "total": 2938 |
| }, |
| "real word": { |
| "right": 0.1168, |
| "wrong": 0.0075, |
| "total": 1062 |
| }, |
| "all right": 0.5357, |
| "all wrong": 0.0395, |
| "email changed per 1000": 0.966, |
| "github changed per 1000": 2.199 |
| } |
| } |
| } |
| }, |
| "sizes_mb": { |
| "matmul-gather": 2.15, |
| "matmul": 6.05 |
| }, |
| "shipped": "matmul-gather" |
| } |
|
|
