Model update

README.md

@@ -1,5 +1,5 @@
 ---
-title: Test2
+title: AMRdemo
 emoji: 😻
 colorFrom: indigo
 colorTo: blue

model_stog/amrlib_meta.json

@@ -5,6 +5,6 @@
     "inference_module":".parse_xfm.inference", 
     "inference_class":"Inference",
     "model_fn":"pytorch_model.bin",
-    "base_model":"facebook/bart-large",
+    "base_model":"facebook/bart-base",
     "kwargs":{}
 }

model_stog/config.json

@@ -1,5 +1,5 @@
 {
-  "_name_or_path": "facebook/bart-large",
+  "_name_or_path": "facebook/bart-base",
   "activation_dropout": 0.1,
   "activation_function": "gelu",
   "add_bias_logits": false,
@@ -11,18 +11,18 @@
   "bos_token_id": 0,
   "classif_dropout": 0.1,
   "classifier_dropout": 0.0,
-  "d_model": 1024,
-  "decoder_attention_heads": 16,
-  "decoder_ffn_dim": 4096,
+  "d_model": 768,
+  "decoder_attention_heads": 12,
+  "decoder_ffn_dim": 3072,
   "decoder_layerdrop": 0.0,
-  "decoder_layers": 12,
+  "decoder_layers": 6,
   "decoder_start_token_id": 2,
   "dropout": 0.1,
   "early_stopping": true,
-  "encoder_attention_heads": 16,
-  "encoder_ffn_dim": 4096,
+  "encoder_attention_heads": 12,
+  "encoder_ffn_dim": 3072,
   "encoder_layerdrop": 0.0,
-  "encoder_layers": 12,
+  "encoder_layers": 6,
   "eos_token_id": 2,
   "forced_eos_token_id": 2,
   "gradient_checkpointing": false,
@@ -41,8 +41,9 @@
   "max_position_embeddings": 1024,
   "model_type": "bart",
   "normalize_before": false,
+  "normalize_embedding": true,
   "num_beams": 4,
-  "num_hidden_layers": 12,
+  "num_hidden_layers": 6,
   "pad_token_id": 1,
   "scale_embedding": false,
   "task_specific_params": {
@@ -57,7 +58,7 @@
       "max_out_len": 1024,
       "max_train_graph_len": 512,
       "max_train_sent_len": 100,
-      "model_name_or_path": "facebook/bart-large",
+      "model_name_or_path": "facebook/bart-base",
       "save_at_end": false,
       "save_tokenizer": false,
       "train_fn": "train.txt.nowiki"

model_stog/{model_parse_xfm_bart_large.json → model_parse_xfm_bart_base.json}

@@ -1,6 +1,6 @@
 {   "gen_args" :
     {
-        "model_name_or_path"            : "facebook/bart-large",
+        "model_name_or_path"            : "facebook/bart-base",
         "corpus_dir"                    : "data/tdata_xfm/",
         "train_fn"                      : "train.txt.nowiki",
         "eval_fn"                       : "dev.txt.nowiki",
@@ -31,12 +31,12 @@
         "save_total_limit"              : 1,
         "logging_steps"                 : 300,
         "num_train_epochs"              : 16,
-        "per_device_train_batch_size"   : 8,
-        "gradient_accumulation_steps"   : 4,
+        "per_device_train_batch_size"   : 16,
+        "gradient_accumulation_steps"   : 1,
         "weight_decay"                  : 0.004,
         "learning_rate"                 : 5e-5,
         "max_grad_norm"                 : 1.0,
-        "warmup_steps"                  : 5200,
+        "warmup_steps"                  : 3467,
         "seed"                          : 42
     }
 }

model_stog/pytorch_model.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:d9bdcde696e63224ba56853469689d2e22e64c324eedee6e48348e806eedd45b
-size 1625557313
+oid sha256:281b14c587df85226e182dd220535f5eb5e1a4a499836bc6cfe812f5e0adc5e0
+size 557979193