Upload model

config.json

@@ -8,7 +8,7 @@
     "cls": {
       "combined_dim": 128,
       "combined_input": 384,
-      "n_layers": 8,
+      "n_layers": 4,
       "num_classes": 1,
       "skip_in": [
         80
@@ -16,19 +16,19 @@
       "weight_norm": true
     },
     "freq": {
-      "n_layers": 8,
+      "n_layers": 4,
       "skip_in": [
         80
       ],
       "tfidf_hidden_dim": 512,
-      "tfidf_input_dim": 8192,
+      "tfidf_input_dim": 4096,
       "tfidf_output_dim": 128,
       "weight_norm": true
     },
     "pos": {
       "hidden_dim": 256,
       "input_dim": 512,
-      "n_layers": 8,
+      "n_layers": 4,
       "output_dim": 128,
       "skip_in": [
         80
@@ -40,7 +40,7 @@
       "hidden_dim": 256,
       "input_dim": 768,
       "lstm_in": 768,
-      "n_layers": 8,
+      "n_layers": 4,
       "output_dim": 128,
       "skip_in": [
         80
@@ -53,12 +53,12 @@
   "torch_dtype": "float32",
   "train": {
     "anneal_end": 5,
-    "batch_size": 32,
+    "batch_size": 64,
     "end_iter": 10,
     "learning_rate": 0.0005,
     "learning_rate_alpha": 0.05,
     "save_freq": 5,
-    "val_freq": 2,
+    "val_freq": 1,
     "warm_up_end": 2
   },
   "transformers_version": "4.47.0"

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:6ae66293d423677e46b265e376281c32d6190f8b8a56e445c961e5ab32293639
-size 487360928
+oid sha256:4cf79dabdf1081c08a5fdf3172b6f2300b46325962b2b3e93b0f1cd754a841f1
+size 472382064