Training in progress, step 1000

config.json

@@ -160,14 +160,14 @@
   "label_smoothing": 0.0,
   "length_penalty": 1.0,
   "llm_dim": 2048,
-  "lora_alpha": 32,
-  "lora_dropout": 0.0,
-  "lora_r": 64,
-  "lora_target_modules": [
-    "v_proj",
-    "q_proj"
-  ],
+  "mask_feature_length": 10,
+  "mask_feature_min_masks": 0,
+  "mask_feature_prob": 0.0,
+  "mask_time_length": 10,
+  "mask_time_min_masks": 2,
+  "mask_time_prob": 0.05,
   "max_new_tokens": 256,
+  "min_new_tokens": 1,
   "model_dtype": "bfloat16",
   "model_type": "asr_model",
   "no_repeat_ngram_size": 0,
@@ -178,7 +178,6 @@
   "projector_dropout": 0.0,
   "projector_hidden_dim": null,
   "projector_init_std": 0.02,
-  "projector_input_noise": 0.0,
   "projector_num_layers": 2,
   "projector_pool_stride": 4,
   "projector_type": "mosa",
@@ -255,7 +254,7 @@
   "text_model_id": "Qwen/Qwen3-1.7B",
   "transformers_version": "5.0.0.dev0",
   "use_cache": false,
-  "use_lora": true,
+  "use_lora": false,
   "use_specaugment": true,
   "user_prompt": "Please transcribe this English audio into text: <audio>",
   "vocab_size": 151670

generation_config.json

@@ -6,6 +6,7 @@
   ],
   "length_penalty": 1.0,
   "max_new_tokens": 256,
+  "min_new_tokens": 1,
   "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 sha256:0ca080ca5928f40d46600cd7a02ca2ee693894c41bd91ad8b99d852ae923f118
+oid sha256:1c2a286f15a803944ee979473fd7aeca3cf83242d2541114b5b425cb43158a52
 size 236248080

training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:0ecb28a40b3e5b48c6e2923003e47d334c96cb66c84c3f24cf7192e4e93c361c
+oid sha256:33e9003ce34c401f33cf18981a0f24a8877c856874f33f16e65e105a4f86f557
 size 5201