| { |
| "iterations": 2400, |
| "parameter_count": 120349440, |
| "tokens_per_step": 131072, |
| "processed_tokens": 314572800, |
| "elapsed_seconds": 5886.268626770005, |
| "best_validation_loss": 5.134598255157471, |
| "history": [ |
| { |
| "iteration": 100, |
| "train_loss": 7.84290885925293, |
| "validation_loss": 7.806333065032959, |
| "learning_rate": 0.0005939999999999999, |
| "elapsed_seconds": 236.28617264330387, |
| "projected_total_hours": 0.3281752397823665, |
| "projected_total_cost_usd": 0.08860731474123897 |
| }, |
| { |
| "iteration": 200, |
| "train_loss": 7.040276050567627, |
| "validation_loss": 7.100771427154541, |
| "learning_rate": 0.0005942168441678774, |
| "elapsed_seconds": 480.88143431581557, |
| "projected_total_hours": 0.3339454404970941, |
| "projected_total_cost_usd": 0.09016526893421542 |
| }, |
| { |
| "iteration": 300, |
| "train_loss": 6.685205459594727, |
| "validation_loss": 6.788591384887695, |
| "learning_rate": 0.0005768867365685819, |
| "elapsed_seconds": 727.3782979361713, |
| "projected_total_hours": 0.3367492120074867, |
| "projected_total_cost_usd": 0.09092228724202142 |
| }, |
| { |
| "iteration": 400, |
| "train_loss": 6.431410312652588, |
| "validation_loss": 6.551764011383057, |
| "learning_rate": 0.0005487664938872375, |
| "elapsed_seconds": 971.8389842156321, |
| "projected_total_hours": 0.33744409174153894, |
| "projected_total_cost_usd": 0.09110990477021552 |
| }, |
| { |
| "iteration": 500, |
| "train_loss": 6.205532073974609, |
| "validation_loss": 6.343863487243652, |
| "learning_rate": 0.0005110851056649143, |
| "elapsed_seconds": 1216.9689059387892, |
| "projected_total_hours": 0.3380469183163303, |
| "projected_total_cost_usd": 0.0912726679454092 |
| }, |
| { |
| "iteration": 600, |
| "train_loss": 6.005918979644775, |
| "validation_loss": 6.159409046173096, |
| "learning_rate": 0.0004654894293822898, |
| "elapsed_seconds": 1464.3989623263478, |
| "projected_total_hours": 1.084739972093591, |
| "projected_total_cost_usd": 0.29287979246526963 |
| }, |
| { |
| "iteration": 700, |
| "train_loss": 5.821047306060791, |
| "validation_loss": 5.991540431976318, |
| "learning_rate": 0.000413972214885244, |
| "elapsed_seconds": 1711.4358539339155, |
| "projected_total_hours": 1.0866259390056607, |
| "projected_total_cost_usd": 0.2933890035315284 |
| }, |
| { |
| "iteration": 800, |
| "train_loss": 5.646799087524414, |
| "validation_loss": 5.839986801147461, |
| "learning_rate": 0.0003587850116543201, |
| "elapsed_seconds": 1958.1738921981305, |
| "projected_total_hours": 1.087874384554517, |
| "projected_total_cost_usd": 0.2937260838297196 |
| }, |
| { |
| "iteration": 900, |
| "train_loss": 5.511969566345215, |
| "validation_loss": 5.7192158699035645, |
| "learning_rate": 0.0003023397652882918, |
| "elapsed_seconds": 2205.4278924595565, |
| "projected_total_hours": 1.0891001938071883, |
| "projected_total_cost_usd": 0.2940570523279408 |
| }, |
| { |
| "iteration": 1000, |
| "train_loss": 5.387481689453125, |
| "validation_loss": 5.612607479095459, |
| "learning_rate": 0.00024710340390769715, |
| "elapsed_seconds": 2451.299042198807, |
| "projected_total_hours": 1.0894662409772475, |
| "projected_total_cost_usd": 0.29415588506385687 |
| }, |
| { |
| "iteration": 1100, |
| "train_loss": 5.2906174659729, |
| "validation_loss": 5.521340847015381, |
| "learning_rate": 0.00019549002155833746, |
| "elapsed_seconds": 2694.78519503586, |
| "projected_total_hours": 1.0888020990043878, |
| "projected_total_cost_usd": 0.2939765667311847 |
| }, |
| { |
| "iteration": 1200, |
| "train_loss": 5.1990485191345215, |
| "validation_loss": 5.441063404083252, |
| "learning_rate": 0.00014975537071736653, |
| "elapsed_seconds": 2938.766986746341, |
| "projected_total_hours": 1.0884322173134595, |
| "projected_total_cost_usd": 0.29387669867463406 |
| }, |
| { |
| "iteration": 1300, |
| "train_loss": 5.1335320472717285, |
| "validation_loss": 5.377827167510986, |
| "learning_rate": 0.00011189827508573826, |
| "elapsed_seconds": 3183.2163684107363, |
| "projected_total_hours": 1.0882791003113628, |
| "projected_total_cost_usd": 0.293835357084068 |
| }, |
| { |
| "iteration": 1400, |
| "train_loss": 5.080015182495117, |
| "validation_loss": 5.331523895263672, |
| "learning_rate": 8.357327140105425e-05, |
| "elapsed_seconds": 3428.121685290709, |
| "projected_total_hours": 1.088292598504987, |
| "projected_total_cost_usd": 0.2938390015963465 |
| }, |
| { |
| "iteration": 1500, |
| "train_loss": 5.044103622436523, |
| "validation_loss": 5.2975358963012695, |
| "learning_rate": 6.601829824278278e-05, |
| "elapsed_seconds": 3671.2561981901526, |
| "projected_total_hours": 1.0877796142785638, |
| "projected_total_cost_usd": 0.29370049585521224 |
| }, |
| { |
| "iteration": 1600, |
| "train_loss": 5.01863956451416, |
| "validation_loss": 5.270280361175537, |
| "learning_rate": 6.0000592176047586e-05, |
| "elapsed_seconds": 3915.9751346502453, |
| "projected_total_hours": 1.0877708707361793, |
| "projected_total_cost_usd": 0.2936981350987684 |
| }, |
| { |
| "iteration": 1700, |
| "train_loss": 4.996570587158203, |
| "validation_loss": 5.251531600952148, |
| "learning_rate": 6e-05, |
| "elapsed_seconds": 4161.210727300495, |
| "projected_total_hours": 1.63184734403941, |
| "projected_total_cost_usd": 0.4405987828906407 |
| }, |
| { |
| "iteration": 1800, |
| "train_loss": 4.976433277130127, |
| "validation_loss": 5.230465888977051, |
| "learning_rate": 6e-05, |
| "elapsed_seconds": 4407.253906426951, |
| "projected_total_hours": 1.6323162616396116, |
| "projected_total_cost_usd": 0.44072539064269517 |
| }, |
| { |
| "iteration": 1900, |
| "train_loss": 4.959025859832764, |
| "validation_loss": 5.213982105255127, |
| "learning_rate": 6e-05, |
| "elapsed_seconds": 4654.531830655411, |
| "projected_total_hours": 1.6331690633878635, |
| "projected_total_cost_usd": 0.4409556471147232 |
| }, |
| { |
| "iteration": 2000, |
| "train_loss": 4.941042900085449, |
| "validation_loss": 5.198922157287598, |
| "learning_rate": 6e-05, |
| "elapsed_seconds": 4899.879536379129, |
| "projected_total_hours": 1.633293178793043, |
| "projected_total_cost_usd": 0.4409891582741216 |
| }, |
| { |
| "iteration": 2100, |
| "train_loss": 4.923161506652832, |
| "validation_loss": 5.181520462036133, |
| "learning_rate": 6e-05, |
| "elapsed_seconds": 5145.104391111061, |
| "projected_total_hours": 1.6333664733685909, |
| "projected_total_cost_usd": 0.44100894780951955 |
| }, |
| { |
| "iteration": 2200, |
| "train_loss": 4.907833099365234, |
| "validation_loss": 5.165126323699951, |
| "learning_rate": 6e-05, |
| "elapsed_seconds": 5389.233639188111, |
| "projected_total_hours": 1.6331011027842761, |
| "projected_total_cost_usd": 0.44093729775175455 |
| }, |
| { |
| "iteration": 2300, |
| "train_loss": 4.896564483642578, |
| "validation_loss": 5.152801513671875, |
| "learning_rate": 6e-05, |
| "elapsed_seconds": 5632.714369047433, |
| "projected_total_hours": 1.632670831607952, |
| "projected_total_cost_usd": 0.44082112453414707 |
| }, |
| { |
| "iteration": 2400, |
| "train_loss": 4.876748561859131, |
| "validation_loss": 5.134598255157471, |
| "learning_rate": 6e-05, |
| "elapsed_seconds": 5877.340094184503, |
| "projected_total_hours": 1.6325944706068063, |
| "projected_total_cost_usd": 0.4408005070638377 |
| } |
| ], |
| "environment": { |
| "python": "3.11.10", |
| "platform": "Linux-5.15.0-107-generic-x86_64-with-glibc2.35", |
| "torch": "2.5.1+cu124", |
| "cuda": "12.4", |
| "gpu": "NVIDIA RTX A5000" |
| }, |
| "data_manifest_sha256": "f492224f8bc7083fdac3c15ed348b5b0f627436a5e26c4fff109b3a6fe3f06de", |
| "peak_vram_bytes": 4921769984, |
| "gpu_total_bytes": 25293946880, |
| "estimated_training_cost_usd": 0.4414701470077504 |
| } |
|
|
