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/30k",
-    "log_file": "/kaggle/working/logs/finetune_hi_30k.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": 30000,
-    "epochs": 3.0,
-    "max_steps": 0,
-    "eval_every": 140,
-    "ckpt_every": 702,
-    "log_every": 25,
-    "device": "cuda",
-    "seed": 1337,
-    "marker": "जवाब"
-  },
-  "pretrained_step": 6998,
-  "pretrained_val_loss": 3.247800827026367,
-  "train_examples": 30000,
-  "val_examples": 2000,
-  "train_limit": 30000,
-  "steps": 2811,
-  "train_seconds": 110.6,
-  "best_val_loss": 0.9830271418278034,
-  "history": [
-    {
-      "step": 140,
-      "train_loss": 1.4107627868652344,
-      "val_loss": 1.3869942610080426
-    },
-    {
-      "step": 280,
-      "train_loss": 1.1072734594345093,
-      "val_loss": 1.2451391220092773
-    },
-    {
-      "step": 420,
-      "train_loss": 1.2033661603927612,
-      "val_loss": 1.1584641612493074
-    },
-    {
-      "step": 560,
-      "train_loss": 1.184777855873108,
-      "val_loss": 1.161140643633329
-    },
-    {
-      "step": 700,
-      "train_loss": 1.0675430297851562,
-      "val_loss": 1.0932962160844069
-    },
-    {
-      "step": 840,
-      "train_loss": 1.093224287033081,
-      "val_loss": 1.0843981641989489
-    },
-    {
-      "step": 980,
-      "train_loss": 1.027336597442627,
-      "val_loss": 1.1282126674285302
-    },
-    {
-      "step": 1120,
-      "train_loss": 0.9259868264198303,
-      "val_loss": 1.0795260163453908
-    },
-    {
-      "step": 1260,
-      "train_loss": 1.0070079565048218,
-      "val_loss": 1.107613769861368
-    },
-    {
-      "step": 1400,
-      "train_loss": 0.906167209148407,
-      "val_loss": 1.085576621385721
-    },
-    {
-      "step": 1540,
-      "train_loss": 0.8806331157684326,
-      "val_loss": 1.0580799763019269
-    },
-    {
-      "step": 1680,
-      "train_loss": 1.0024728775024414,
-      "val_loss": 1.0651229895078218
-    },
-    {
-      "step": 1820,
-      "train_loss": 0.9697971343994141,
-      "val_loss": 0.9946631284860464
-    },
-    {
-      "step": 1960,
-      "train_loss": 1.0033124685287476,
-      "val_loss": 1.0303042989510756
-    },
-    {
-      "step": 2100,
-      "train_loss": 0.9343948364257812,
-      "val_loss": 1.0205647807854872
-    },
-    {
-      "step": 2240,
-      "train_loss": 0.975424587726593,
-      "val_loss": 0.9954107724703275
-    },
-    {
-      "step": 2380,
-      "train_loss": 0.8892805576324463,
-      "val_loss": 0.9968066536463224
-    },
-    {
-      "step": 2520,
-      "train_loss": 0.785287618637085,
-      "val_loss": 0.9830271418278034
-    },
-    {
-      "step": 2660,
-      "train_loss": 0.8383352756500244,
-      "val_loss": 0.9983221750992995
-    },
-    {
-      "step": 2800,
-      "train_loss": 0.844124436378479,
-      "val_loss": 0.9888775394513056
-    },
-    {
-      "step": 2811,
-      "train_loss": 0.9225854873657227,
-      "val_loss": 0.9874970133487995
-    }
-  ]
-}