Training in progress, step 1000

config.json

@@ -160,14 +160,14 @@
   "label_smoothing": 0.0,
   "length_penalty": 1.0,
   "llm_dim": 2048,
-  "mask_feature_length": 10,
-  "mask_feature_min_masks": 0,
-  "mask_feature_prob": 0.0,
-  "mask_time_length": 10,
+  "mask_feature_length": 27,
+  "mask_feature_min_masks": 2,
+  "mask_feature_prob": 1.0,
+  "mask_time_length": 100,
   "mask_time_min_masks": 2,
   "mask_time_prob": 0.05,
   "max_new_tokens": 256,
-  "min_new_tokens": 1,
+  "min_new_tokens": 0,
   "model_dtype": "bfloat16",
   "model_type": "asr_model",
   "no_repeat_ngram_size": 0,
@@ -179,7 +179,7 @@
   "projector_hidden_dim": null,
   "projector_init_std": 0.02,
   "projector_num_layers": 2,
-  "projector_pool_stride": 4,
+  "projector_pool_stride": 2,
   "projector_type": "mosa",
   "qformer_hidden_size": null,
   "qformer_intermediate_size": null,

generation_config.json

@@ -6,7 +6,7 @@
   ],
   "length_penalty": 1.0,
   "max_new_tokens": 256,
-  "min_new_tokens": 1,
+  "min_new_tokens": 0,
   "no_repeat_ngram_size": 0,
   "num_beams": 1,
   "pad_token_id": 151643,

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:4a1c89fa6a1ceb6e1fb602691b46f27457211305a935779e465dc53e875cda81
-size 236248080
+oid sha256:647480fde66d2118237a959475e29edaf88be4659e13c6f270662a4764708b9e
+size 152361992

training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:cffa015a74c99c00afa442e17b9016702915c3b296117c2edd7cb23230cb711d
+oid sha256:bc08df300569bac0c1ca05ca48ecacdd1f1b95aced10687606033a87f3f08720
 size 5201