meet5568
/

lma_models

@@ -1,154 +0,0 @@
-{
-  "model": {
-    "lang": "hi",
-    "name": "model_h",
-    "vocab_size": 10000,
-    "d_model": 640,
-    "n_layers": 4,
-    "n_heads": 8,
-    "d_ff": 2560,
-    "context_length": 512,
-    "dropout": 0.0,
-    "positional": "learned",
-    "tie_embeddings": true,
-    "bias": true
-  },
-  "finetune": {
-    "lang": "hi",
-    "data_dir": "/kaggle/input/datasets/meet6868/lma-ft-hi",
-    "init_from": "/kaggle/input/datasets/meet6868/lma-ft-hi/pretrained_best.pt",
-    "tokenizer": "/kaggle/input/datasets/meet6868/lma-ft-hi/hi_tokenizer.model",
-    "out_dir": "/kaggle/working/sweep/50k",
-    "log_file": "/kaggle/working/logs/finetune_hi_50k.log",
-    "peak_lr": 0.0001,
-    "min_lr": 1e-05,
-    "warmup_ratio": 0.05,
-    "weight_decay": 0.1,
-    "grad_clip": 1.0,
-    "batch_size": 32,
-    "train_limit": 50000,
-    "epochs": 3.0,
-    "max_steps": 0,
-    "eval_every": 234,
-    "ckpt_every": 1171,
-    "log_every": 25,
-    "device": "cuda",
-    "seed": 1337,
-    "marker": "जवाब"
-  },
-  "pretrained_step": 6998,
-  "pretrained_val_loss": 3.247800827026367,
-  "train_examples": 50000,
-  "val_examples": 2000,
-  "train_limit": 50000,
-  "steps": 4686,
-  "train_seconds": 184.3,
-  "best_val_loss": 0.9632859184191778,
-  "history": [
-    {
-      "step": 234,
-      "train_loss": 1.1960666179656982,
-      "val_loss": 1.2904538282981286
-    },
-    {
-      "step": 468,
-      "train_loss": 1.1585614681243896,
-      "val_loss": 1.2431558966636658
-    },
-    {
-      "step": 702,
-      "train_loss": 1.0431594848632812,
-      "val_loss": 1.1556175305293157
-    },
-    {
-      "step": 936,
-      "train_loss": 1.0920562744140625,
-      "val_loss": 1.1721054315567017
-    },
-    {
-      "step": 1170,
-      "train_loss": 1.3762054443359375,
-      "val_loss": 1.1005728198931768
-    },
-    {
-      "step": 1404,
-      "train_loss": 1.4893150329589844,
-      "val_loss": 1.1180422810407786
-    },
-    {
-      "step": 1638,
-      "train_loss": 1.0931141376495361,
-      "val_loss": 1.0475326363856976
-    },
-    {
-      "step": 1872,
-      "train_loss": 1.0343246459960938,
-      "val_loss": 1.0573025804299574
-    },
-    {
-      "step": 2106,
-      "train_loss": 1.1081695556640625,
-      "val_loss": 1.0493067457125738
-    },
-    {
-      "step": 2340,
-      "train_loss": 1.07891845703125,
-      "val_loss": 1.0647912896596468
-    },
-    {
-      "step": 2574,
-      "train_loss": 0.9994146227836609,
-      "val_loss": 1.0690545302170973
-    },
-    {
-      "step": 2808,
-      "train_loss": 0.9543197751045227,
-      "val_loss": 1.0516250683711126
-    },
-    {
-      "step": 3042,
-      "train_loss": 0.8230546712875366,
-      "val_loss": 1.0328002709608812
-    },
-    {
-      "step": 3276,
-      "train_loss": 0.8311129808425903,
-      "val_loss": 1.0057905224653392
-    },
-    {
-      "step": 3510,
-      "train_loss": 0.8408849835395813,
-      "val_loss": 1.0086686244377723
-    },
-    {
-      "step": 3744,
-      "train_loss": 0.8777179718017578,
-      "val_loss": 0.987655873482044
-    },
-    {
-      "step": 3978,
-      "train_loss": 1.0006103515625,
-      "val_loss": 0.9933772866542523
-    },
-    {
-      "step": 4212,
-      "train_loss": 1.0678176879882812,
-      "val_loss": 0.9632859184191778
-    },
-    {
-      "step": 4446,
-      "train_loss": 0.7339738011360168,
-      "val_loss": 0.9799387775934659
-    },
-    {
-      "step": 4680,
-      "train_loss": 0.9961695671081543,
-      "val_loss": 0.9985149273505578
-    },
-    {
-      "step": 4686,
-      "train_loss": 0.81858891248703,
-      "val_loss": 0.9990851237223699
-    }
-  ]
-}