Training in progress, epoch 0

config.json

@@ -1,16 +1,16 @@
 {
-  "_name_or_path": "MCG-NJU/videomae-base",
+  "_name_or_path": "MCG-NJU/videomae-small-finetuned-ssv2",
   "architectures": [
     "VideoMAEForVideoClassification"
   ],
   "attention_probs_dropout_prob": 0.0,
-  "decoder_hidden_size": 384,
-  "decoder_intermediate_size": 1536,
-  "decoder_num_attention_heads": 6,
-  "decoder_num_hidden_layers": 4,
+  "decoder_hidden_size": 192,
+  "decoder_intermediate_size": 768,
+  "decoder_num_attention_heads": 3,
+  "decoder_num_hidden_layers": 12,
   "hidden_act": "gelu",
   "hidden_dropout_prob": 0.0,
-  "hidden_size": 768,
+  "hidden_size": 384,
   "id2label": {
     "0": "furto",
     "1": "normal",
@@ -18,7 +18,7 @@
   },
   "image_size": 224,
   "initializer_range": 0.02,
-  "intermediate_size": 3072,
+  "intermediate_size": 1536,
   "label2id": {
     "furto": 0,
     "normal": 1,
@@ -27,7 +27,7 @@
   "layer_norm_eps": 1e-12,
   "model_type": "videomae",
   "norm_pix_loss": true,
-  "num_attention_heads": 12,
+  "num_attention_heads": 16,
   "num_channels": 3,
   "num_frames": 15,
   "num_hidden_layers": 12,
@@ -37,5 +37,5 @@
   "torch_dtype": "float32",
   "transformers_version": "4.44.2",
   "tubelet_size": 2,
-  "use_mean_pooling": false
+  "use_mean_pooling": true
 }

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:2177768e567e13c16fd76a7f02867f51b540295bafa64f4121929dc10aa9cfa4
-size 344940444
+oid sha256:f493472702bab7723e5848b41f55e230749f5efff98a3a115cfa0dfb3026a065
+size 87546468

training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:3ca48d3ee58c152fc4a0fb22e94f6448b11ecba3a26568507682ea63ef5c20e5
+oid sha256:3e2be7bc9e667279a4c01c57bce9826bab1c0f5677c39c1fb0dd4ca3b411f0a6
 size 5176