Upload folder using huggingface_hub

.gitattributes

@@ -33,3 +33,4 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
 *.zip filter=lfs diff=lfs merge=lfs -text
 *.zst filter=lfs diff=lfs merge=lfs -text
 *tfevents* filter=lfs diff=lfs merge=lfs -text
+tokenizer.json filter=lfs diff=lfs merge=lfs -text

.gitignore

@@ -0,0 +1,4 @@
+*.pt
+*.pth
+*.bin
+*.txt

added_tokens.json

@@ -0,0 +1,26 @@
+{
+  "</tool_call>": 151658,
+  "<tool_call>": 151657,
+  "<|box_end|>": 151649,
+  "<|box_start|>": 151648,
+  "<|end_of_audio|>": 151666,
+  "<|endoftext|>": 151643,
+  "<|file_sep|>": 151664,
+  "<|fim_middle|>": 151660,
+  "<|fim_pad|>": 151662,
+  "<|fim_prefix|>": 151659,
+  "<|fim_suffix|>": 151661,
+  "<|im_end|>": 151645,
+  "<|im_start|>": 151644,
+  "<|image_pad|>": 151655,
+  "<|object_ref_end|>": 151647,
+  "<|object_ref_start|>": 151646,
+  "<|quad_end|>": 151651,
+  "<|quad_start|>": 151650,
+  "<|repo_name|>": 151663,
+  "<|start_of_audio|>": 151665,
+  "<|video_pad|>": 151656,
+  "<|vision_end|>": 151653,
+  "<|vision_pad|>": 151654,
+  "<|vision_start|>": 151652
+}

config.json

@@ -0,0 +1,30 @@
+{
+  "_name_or_path": "ksych/salt-asr",
+  "architectures": [
+    "Qwen2ForCausalLM"
+  ],
+  "attention_dropout": 0.0,
+  "bos_token_id": 151643,
+  "eos_token_id": 151643,
+  "hidden_act": "silu",
+  "hidden_size": 896,
+  "initializer_range": 0.02,
+  "intermediate_size": 4864,
+  "max_position_embeddings": 32768,
+  "max_window_layers": 24,
+  "model_type": "qwen2",
+  "num_attention_heads": 14,
+  "num_hidden_layers": 24,
+  "num_key_value_heads": 2,
+  "rms_norm_eps": 1e-06,
+  "rope_scaling": null,
+  "rope_theta": 1000000.0,
+  "sliding_window": 32768,
+  "tie_word_embeddings": true,
+  "torch_dtype": "float32",
+  "transformers_version": "4.49.0",
+  "use_cache": false,
+  "use_mrope": false,
+  "use_sliding_window": false,
+  "vocab_size": 160883
+}

generation_config.json

@@ -0,0 +1,6 @@
+{
+  "bos_token_id": 151643,
+  "eos_token_id": 151643,
+  "max_new_tokens": 2048,
+  "transformers_version": "4.49.0"
+}

model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:df1da6236b5cbd0f91fc6d79f9f9c74e1d5803f3c9ab5a9b1f19828756c9cd05
+size 2008229528

special_tokens_map.json

@@ -0,0 +1,32 @@
+{
+  "additional_special_tokens": [
+    {
+      "content": "<|start_of_audio|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false
+    },
+    {
+      "content": "<|end_of_audio|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false
+    }
+  ],
+  "eos_token": {
+    "content": "<|endoftext|>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "pad_token": {
+    "content": "<|endoftext|>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  }
+}

tokenizer.json

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:55c7fad3b807310f01cead0edd8fa225070d199053eb0649e31f58a1caf09aa2
+size 11422284

tokenizer_config.json

@@ -0,0 +1,213 @@
+{
+  "add_bos_token": false,
+  "add_prefix_space": false,
+  "added_tokens_decoder": {
+    "151643": {
+      "content": "<|endoftext|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151644": {
+      "content": "<|im_start|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151645": {
+      "content": "<|im_end|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151646": {
+      "content": "<|object_ref_start|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151647": {
+      "content": "<|object_ref_end|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151648": {
+      "content": "<|box_start|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151649": {
+      "content": "<|box_end|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151650": {
+      "content": "<|quad_start|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151651": {
+      "content": "<|quad_end|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151652": {
+      "content": "<|vision_start|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151653": {
+      "content": "<|vision_end|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151654": {
+      "content": "<|vision_pad|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151655": {
+      "content": "<|image_pad|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151656": {
+      "content": "<|video_pad|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151657": {
+      "content": "<tool_call>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151658": {
+      "content": "</tool_call>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151659": {
+      "content": "<|fim_prefix|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151660": {
+      "content": "<|fim_middle|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151661": {
+      "content": "<|fim_suffix|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151662": {
+      "content": "<|fim_pad|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151663": {
+      "content": "<|repo_name|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151664": {
+      "content": "<|file_sep|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151665": {
+      "content": "<|start_of_audio|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151666": {
+      "content": "<|end_of_audio|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    }
+  },
+  "additional_special_tokens": [
+    "<|start_of_audio|>",
+    "<|end_of_audio|>"
+  ],
+  "bos_token": null,
+  "chat_template": "{%- if tools %}\n    {{- '<|im_start|>system\\n' }}\n    {%- if messages[0]['role'] == 'system' %}\n        {{- messages[0]['content'] }}\n    {%- else %}\n        {{- 'You are a helpful assistant.' }}\n    {%- endif %}\n    {{- \"\\n\\n# Tools\\n\\nYou may call one or more functions to assist with the user query.\\n\\nYou are provided with function signatures within <tools></tools> XML tags:\\n<tools>\" }}\n    {%- for tool in tools %}\n        {{- \"\\n\" }}\n        {{- tool | tojson }}\n    {%- endfor %}\n    {{- \"\\n</tools>\\n\\nFor each function call, return a json object with function name and arguments within <tool_call></tool_call> XML tags:\\n<tool_call>\\n{\\\"name\\\": <function-name>, \\\"arguments\\\": <args-json-object>}\\n</tool_call><|im_end|>\\n\" }}\n{%- else %}\n    {%- if messages[0]['role'] == 'system' %}\n        {{- '<|im_start|>system\\n' + messages[0]['content'] + '<|im_end|>\\n' }}\n    {%- else %}\n        {{- '<|im_start|>system\\nYou are a helpful assistant.<|im_end|>\\n' }}\n    {%- endif %}\n{%- endif %}\n{%- for message in messages %}\n    {%- if (message.role == \"user\") or (message.role == \"system\" and not loop.first) or (message.role == \"assistant\" and not message.tool_calls) %}\n        {{- '<|im_start|>' + message.role + '\\n' + message.content + '<|im_end|>' + '\\n' }}\n    {%- elif message.role == \"assistant\" %}\n        {{- '<|im_start|>' + message.role }}\n        {%- if message.content %}\n            {{- '\\n' + message.content }}\n        {%- endif %}\n        {%- for tool_call in message.tool_calls %}\n            {%- if tool_call.function is defined %}\n                {%- set tool_call = tool_call.function %}\n            {%- endif %}\n            {{- '\\n<tool_call>\\n{\"name\": \"' }}\n            {{- tool_call.name }}\n            {{- '\", \"arguments\": ' }}\n            {{- tool_call.arguments | tojson }}\n            {{- '}\\n</tool_call>' }}\n        {%- endfor %}\n        {{- '<|im_end|>\\n' }}\n    {%- elif message.role == \"tool\" %}\n        {%- if (loop.index0 == 0) or (messages[loop.index0 - 1].role != \"tool\") %}\n            {{- '<|im_start|>user' }}\n        {%- endif %}\n        {{- '\\n<tool_response>\\n' }}\n        {{- message.content }}\n        {{- '\\n</tool_response>' }}\n        {%- if loop.last or (messages[loop.index0 + 1].role != \"tool\") %}\n            {{- '<|im_end|>\\n' }}\n        {%- endif %}\n    {%- endif %}\n{%- endfor %}\n{%- if add_generation_prompt %}\n    {{- '<|im_start|>assistant\\n' }}\n{%- endif %}\n",
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<|endoftext|>",
+  "errors": "replace",
+  "extra_special_tokens": {},
+  "model_max_length": 131072,
+  "pad_token": "<|endoftext|>",
+  "split_special_tokens": false,
+  "tokenizer_class": "Qwen2Tokenizer",
+  "unk_token": null
+}

trainer_state.json

@@ -0,0 +1,2353 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.33695652173913043,
+  "eval_steps": 1000,
+  "global_step": 15500,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0010869565217391304,
+      "grad_norm": 7.2570319175720215,
+      "learning_rate": 4.5e-06,
+      "loss": 11.9743,
+      "step": 50
+    },
+    {
+      "epoch": 0.002173913043478261,
+      "grad_norm": 40.365055084228516,
+      "learning_rate": 9.5e-06,
+      "loss": 10.4555,
+      "step": 100
+    },
+    {
+      "epoch": 0.003260869565217391,
+      "grad_norm": 12.999434471130371,
+      "learning_rate": 1.44e-05,
+      "loss": 8.1836,
+      "step": 150
+    },
+    {
+      "epoch": 0.004347826086956522,
+      "grad_norm": 17.2613468170166,
+      "learning_rate": 1.94e-05,
+      "loss": 6.3993,
+      "step": 200
+    },
+    {
+      "epoch": 0.005434782608695652,
+      "grad_norm": 9.532617568969727,
+      "learning_rate": 2.44e-05,
+      "loss": 5.8655,
+      "step": 250
+    },
+    {
+      "epoch": 0.006521739130434782,
+      "grad_norm": 4.081784248352051,
+      "learning_rate": 2.94e-05,
+      "loss": 5.2595,
+      "step": 300
+    },
+    {
+      "epoch": 0.007608695652173913,
+      "grad_norm": 4.626275539398193,
+      "learning_rate": 3.4399999999999996e-05,
+      "loss": 5.8076,
+      "step": 350
+    },
+    {
+      "epoch": 0.008695652173913044,
+      "grad_norm": 9.461868286132812,
+      "learning_rate": 3.94e-05,
+      "loss": 5.495,
+      "step": 400
+    },
+    {
+      "epoch": 0.009782608695652175,
+      "grad_norm": 3.323793649673462,
+      "learning_rate": 4.44e-05,
+      "loss": 5.7005,
+      "step": 450
+    },
+    {
+      "epoch": 0.010869565217391304,
+      "grad_norm": 4.26762580871582,
+      "learning_rate": 4.94e-05,
+      "loss": 5.256,
+      "step": 500
+    },
+    {
+      "epoch": 0.011956521739130435,
+      "grad_norm": 3.8675103187561035,
+      "learning_rate": 5.440000000000001e-05,
+      "loss": 5.6752,
+      "step": 550
+    },
+    {
+      "epoch": 0.013043478260869565,
+      "grad_norm": 3.8780524730682373,
+      "learning_rate": 5.94e-05,
+      "loss": 5.0424,
+      "step": 600
+    },
+    {
+      "epoch": 0.014130434782608696,
+      "grad_norm": 5.3671650886535645,
+      "learning_rate": 6.440000000000001e-05,
+      "loss": 4.9488,
+      "step": 650
+    },
+    {
+      "epoch": 0.015217391304347827,
+      "grad_norm": 7.779557228088379,
+      "learning_rate": 6.939999999999999e-05,
+      "loss": 5.0169,
+      "step": 700
+    },
+    {
+      "epoch": 0.016304347826086956,
+      "grad_norm": 5.967152118682861,
+      "learning_rate": 7.44e-05,
+      "loss": 5.2884,
+      "step": 750
+    },
+    {
+      "epoch": 0.017391304347826087,
+      "grad_norm": 5.691137313842773,
+      "learning_rate": 7.94e-05,
+      "loss": 5.4386,
+      "step": 800
+    },
+    {
+      "epoch": 0.01847826086956522,
+      "grad_norm": 7.359672546386719,
+      "learning_rate": 8.44e-05,
+      "loss": 4.932,
+      "step": 850
+    },
+    {
+      "epoch": 0.01956521739130435,
+      "grad_norm": 3.2483417987823486,
+      "learning_rate": 8.94e-05,
+      "loss": 5.3881,
+      "step": 900
+    },
+    {
+      "epoch": 0.020652173913043477,
+      "grad_norm": 4.119468688964844,
+      "learning_rate": 9.44e-05,
+      "loss": 5.0405,
+      "step": 950
+    },
+    {
+      "epoch": 0.021739130434782608,
+      "grad_norm": 5.92612361907959,
+      "learning_rate": 9.94e-05,
+      "loss": 5.2679,
+      "step": 1000
+    },
+    {
+      "epoch": 0.021739130434782608,
+      "eval_asr_loss": 2.527863178171895,
+      "eval_loss": 4.801693916320801,
+      "eval_runtime": 11.3791,
+      "eval_samples_per_second": 56.244,
+      "eval_steps_per_second": 9.403,
+      "eval_tts_loss": 4.963166054594929,
+      "step": 1000
+    },
+    {
+      "epoch": 0.02282608695652174,
+      "grad_norm": 5.158364772796631,
+      "learning_rate": 9.999999089092814e-05,
+      "loss": 5.0704,
+      "step": 1050
+    },
+    {
+      "epoch": 0.02391304347826087,
+      "grad_norm": 4.720920085906982,
+      "learning_rate": 9.999995842574875e-05,
+      "loss": 5.3906,
+      "step": 1100
+    },
+    {
+      "epoch": 0.025,
+      "grad_norm": 3.526841640472412,
+      "learning_rate": 9.999990243509359e-05,
+      "loss": 4.7715,
+      "step": 1150
+    },
+    {
+      "epoch": 0.02608695652173913,
+      "grad_norm": 8.779367446899414,
+      "learning_rate": 9.999982291898898e-05,
+      "loss": 5.4079,
+      "step": 1200
+    },
+    {
+      "epoch": 0.02717391304347826,
+      "grad_norm": 9.412408828735352,
+      "learning_rate": 9.999971987747232e-05,
+      "loss": 5.1509,
+      "step": 1250
+    },
+    {
+      "epoch": 0.02826086956521739,
+      "grad_norm": 7.88346529006958,
+      "learning_rate": 9.999959331059212e-05,
+      "loss": 4.9803,
+      "step": 1300
+    },
+    {
+      "epoch": 0.029347826086956522,
+      "grad_norm": 3.4065446853637695,
+      "learning_rate": 9.999944321840793e-05,
+      "loss": 5.0871,
+      "step": 1350
+    },
+    {
+      "epoch": 0.030434782608695653,
+      "grad_norm": 4.301353931427002,
+      "learning_rate": 9.999926960099035e-05,
+      "loss": 5.1511,
+      "step": 1400
+    },
+    {
+      "epoch": 0.03152173913043478,
+      "grad_norm": 4.603494167327881,
+      "learning_rate": 9.999907245842109e-05,
+      "loss": 4.9227,
+      "step": 1450
+    },
+    {
+      "epoch": 0.03260869565217391,
+      "grad_norm": 7.030359745025635,
+      "learning_rate": 9.999885179079288e-05,
+      "loss": 4.9743,
+      "step": 1500
+    },
+    {
+      "epoch": 0.03369565217391304,
+      "grad_norm": 3.3235089778900146,
+      "learning_rate": 9.999860759820957e-05,
+      "loss": 4.9089,
+      "step": 1550
+    },
+    {
+      "epoch": 0.034782608695652174,
+      "grad_norm": 4.089897632598877,
+      "learning_rate": 9.999833988078603e-05,
+      "loss": 4.9585,
+      "step": 1600
+    },
+    {
+      "epoch": 0.035869565217391305,
+      "grad_norm": 2.7909576892852783,
+      "learning_rate": 9.999804863864826e-05,
+      "loss": 4.7039,
+      "step": 1650
+    },
+    {
+      "epoch": 0.03695652173913044,
+      "grad_norm": 3.653459072113037,
+      "learning_rate": 9.999773387193325e-05,
+      "loss": 4.7419,
+      "step": 1700
+    },
+    {
+      "epoch": 0.03804347826086957,
+      "grad_norm": 2.9489026069641113,
+      "learning_rate": 9.999739558078914e-05,
+      "loss": 4.7499,
+      "step": 1750
+    },
+    {
+      "epoch": 0.0391304347826087,
+      "grad_norm": 3.6647138595581055,
+      "learning_rate": 9.999703376537508e-05,
+      "loss": 4.9439,
+      "step": 1800
+    },
+    {
+      "epoch": 0.04021739130434782,
+      "grad_norm": 7.539992332458496,
+      "learning_rate": 9.999664842586131e-05,
+      "loss": 4.8843,
+      "step": 1850
+    },
+    {
+      "epoch": 0.041304347826086954,
+      "grad_norm": 2.171046257019043,
+      "learning_rate": 9.999623956242915e-05,
+      "loss": 4.7489,
+      "step": 1900
+    },
+    {
+      "epoch": 0.042391304347826085,
+      "grad_norm": 4.13842248916626,
+      "learning_rate": 9.999580717527095e-05,
+      "loss": 4.739,
+      "step": 1950
+    },
+    {
+      "epoch": 0.043478260869565216,
+      "grad_norm": 5.281623363494873,
+      "learning_rate": 9.999535126459016e-05,
+      "loss": 4.9087,
+      "step": 2000
+    },
+    {
+      "epoch": 0.043478260869565216,
+      "eval_asr_loss": 2.592842030251882,
+      "eval_loss": 4.558238506317139,
+      "eval_runtime": 11.4313,
+      "eval_samples_per_second": 55.987,
+      "eval_steps_per_second": 9.36,
+      "eval_tts_loss": 4.98242129726963,
+      "step": 2000
+    },
+    {
+      "epoch": 0.04456521739130435,
+      "grad_norm": 3.798617124557495,
+      "learning_rate": 9.99948718306013e-05,
+      "loss": 4.6324,
+      "step": 2050
+    },
+    {
+      "epoch": 0.04565217391304348,
+      "grad_norm": 3.5986528396606445,
+      "learning_rate": 9.999436887352993e-05,
+      "loss": 4.5383,
+      "step": 2100
+    },
+    {
+      "epoch": 0.04673913043478261,
+      "grad_norm": 2.987210988998413,
+      "learning_rate": 9.999384239361273e-05,
+      "loss": 4.6759,
+      "step": 2150
+    },
+    {
+      "epoch": 0.04782608695652174,
+      "grad_norm": 4.592362880706787,
+      "learning_rate": 9.999329239109738e-05,
+      "loss": 4.7267,
+      "step": 2200
+    },
+    {
+      "epoch": 0.04891304347826087,
+      "grad_norm": 2.9200901985168457,
+      "learning_rate": 9.999271886624266e-05,
+      "loss": 4.5995,
+      "step": 2250
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 3.600618600845337,
+      "learning_rate": 9.999212181931845e-05,
+      "loss": 4.6718,
+      "step": 2300
+    },
+    {
+      "epoch": 0.051086956521739134,
+      "grad_norm": 3.071923017501831,
+      "learning_rate": 9.999150125060565e-05,
+      "loss": 4.791,
+      "step": 2350
+    },
+    {
+      "epoch": 0.05217391304347826,
+      "grad_norm": 3.197481632232666,
+      "learning_rate": 9.999085716039625e-05,
+      "loss": 4.7996,
+      "step": 2400
+    },
+    {
+      "epoch": 0.05326086956521739,
+      "grad_norm": 2.616170883178711,
+      "learning_rate": 9.999018954899328e-05,
+      "loss": 4.7282,
+      "step": 2450
+    },
+    {
+      "epoch": 0.05434782608695652,
+      "grad_norm": 4.689841270446777,
+      "learning_rate": 9.998949841671088e-05,
+      "loss": 4.8002,
+      "step": 2500
+    },
+    {
+      "epoch": 0.05543478260869565,
+      "grad_norm": 2.9560294151306152,
+      "learning_rate": 9.998878376387423e-05,
+      "loss": 4.7618,
+      "step": 2550
+    },
+    {
+      "epoch": 0.05652173913043478,
+      "grad_norm": 5.63021993637085,
+      "learning_rate": 9.998804559081956e-05,
+      "loss": 4.7762,
+      "step": 2600
+    },
+    {
+      "epoch": 0.057608695652173914,
+      "grad_norm": 2.292513370513916,
+      "learning_rate": 9.998728389789422e-05,
+      "loss": 4.5779,
+      "step": 2650
+    },
+    {
+      "epoch": 0.058695652173913045,
+      "grad_norm": 3.1269545555114746,
+      "learning_rate": 9.998649868545659e-05,
+      "loss": 4.5257,
+      "step": 2700
+    },
+    {
+      "epoch": 0.059782608695652176,
+      "grad_norm": 3.462415933609009,
+      "learning_rate": 9.99856899538761e-05,
+      "loss": 4.9268,
+      "step": 2750
+    },
+    {
+      "epoch": 0.06086956521739131,
+      "grad_norm": 2.9171762466430664,
+      "learning_rate": 9.998485770353328e-05,
+      "loss": 4.6967,
+      "step": 2800
+    },
+    {
+      "epoch": 0.06195652173913044,
+      "grad_norm": 10.13646125793457,
+      "learning_rate": 9.998400193481969e-05,
+      "loss": 4.7377,
+      "step": 2850
+    },
+    {
+      "epoch": 0.06304347826086956,
+      "grad_norm": 3.127666711807251,
+      "learning_rate": 9.9983122648138e-05,
+      "loss": 4.612,
+      "step": 2900
+    },
+    {
+      "epoch": 0.0641304347826087,
+      "grad_norm": 4.429073810577393,
+      "learning_rate": 9.998221984390195e-05,
+      "loss": 4.7786,
+      "step": 2950
+    },
+    {
+      "epoch": 0.06521739130434782,
+      "grad_norm": 2.745976686477661,
+      "learning_rate": 9.998129352253626e-05,
+      "loss": 4.7477,
+      "step": 3000
+    },
+    {
+      "epoch": 0.06521739130434782,
+      "eval_asr_loss": 2.4883857901916846,
+      "eval_loss": 4.423598766326904,
+      "eval_runtime": 11.1881,
+      "eval_samples_per_second": 57.204,
+      "eval_steps_per_second": 9.564,
+      "eval_tts_loss": 5.00350919136773,
+      "step": 3000
+    },
+    {
+      "epoch": 0.06630434782608696,
+      "grad_norm": 2.326262950897217,
+      "learning_rate": 9.99803436844768e-05,
+      "loss": 4.5263,
+      "step": 3050
+    },
+    {
+      "epoch": 0.06739130434782609,
+      "grad_norm": 3.688098192214966,
+      "learning_rate": 9.997937033017049e-05,
+      "loss": 4.5735,
+      "step": 3100
+    },
+    {
+      "epoch": 0.06847826086956521,
+      "grad_norm": 2.2472245693206787,
+      "learning_rate": 9.997837346007528e-05,
+      "loss": 4.7224,
+      "step": 3150
+    },
+    {
+      "epoch": 0.06956521739130435,
+      "grad_norm": 1.9073128700256348,
+      "learning_rate": 9.997735307466023e-05,
+      "loss": 4.7719,
+      "step": 3200
+    },
+    {
+      "epoch": 0.07065217391304347,
+      "grad_norm": 3.301027536392212,
+      "learning_rate": 9.997630917440543e-05,
+      "loss": 4.6902,
+      "step": 3250
+    },
+    {
+      "epoch": 0.07173913043478261,
+      "grad_norm": 3.70943021774292,
+      "learning_rate": 9.997524175980204e-05,
+      "loss": 4.8177,
+      "step": 3300
+    },
+    {
+      "epoch": 0.07282608695652174,
+      "grad_norm": 3.3837993144989014,
+      "learning_rate": 9.997415083135229e-05,
+      "loss": 4.8617,
+      "step": 3350
+    },
+    {
+      "epoch": 0.07391304347826087,
+      "grad_norm": 3.145082950592041,
+      "learning_rate": 9.99730363895695e-05,
+      "loss": 4.4194,
+      "step": 3400
+    },
+    {
+      "epoch": 0.075,
+      "grad_norm": 2.647455930709839,
+      "learning_rate": 9.997189843497798e-05,
+      "loss": 4.5414,
+      "step": 3450
+    },
+    {
+      "epoch": 0.07608695652173914,
+      "grad_norm": 2.374641180038452,
+      "learning_rate": 9.997073696811318e-05,
+      "loss": 4.9275,
+      "step": 3500
+    },
+    {
+      "epoch": 0.07717391304347826,
+      "grad_norm": 4.612781047821045,
+      "learning_rate": 9.996955198952159e-05,
+      "loss": 4.9264,
+      "step": 3550
+    },
+    {
+      "epoch": 0.0782608695652174,
+      "grad_norm": 2.460127830505371,
+      "learning_rate": 9.996834349976072e-05,
+      "loss": 4.7562,
+      "step": 3600
+    },
+    {
+      "epoch": 0.07934782608695652,
+      "grad_norm": 2.707995653152466,
+      "learning_rate": 9.996711149939919e-05,
+      "loss": 4.367,
+      "step": 3650
+    },
+    {
+      "epoch": 0.08043478260869565,
+      "grad_norm": 1.578936219215393,
+      "learning_rate": 9.996585598901668e-05,
+      "loss": 4.716,
+      "step": 3700
+    },
+    {
+      "epoch": 0.08152173913043478,
+      "grad_norm": 8.137738227844238,
+      "learning_rate": 9.996457696920392e-05,
+      "loss": 4.5338,
+      "step": 3750
+    },
+    {
+      "epoch": 0.08260869565217391,
+      "grad_norm": 3.5625455379486084,
+      "learning_rate": 9.99632744405627e-05,
+      "loss": 4.2705,
+      "step": 3800
+    },
+    {
+      "epoch": 0.08369565217391305,
+      "grad_norm": 3.893693447113037,
+      "learning_rate": 9.996194840370586e-05,
+      "loss": 4.5428,
+      "step": 3850
+    },
+    {
+      "epoch": 0.08478260869565217,
+      "grad_norm": 3.921764850616455,
+      "learning_rate": 9.996059885925732e-05,
+      "loss": 4.5132,
+      "step": 3900
+    },
+    {
+      "epoch": 0.08586956521739131,
+      "grad_norm": 3.3070826530456543,
+      "learning_rate": 9.995922580785204e-05,
+      "loss": 4.7869,
+      "step": 3950
+    },
+    {
+      "epoch": 0.08695652173913043,
+      "grad_norm": 2.028144598007202,
+      "learning_rate": 9.995782925013609e-05,
+      "loss": 4.5043,
+      "step": 4000
+    },
+    {
+      "epoch": 0.08695652173913043,
+      "eval_asr_loss": 2.603556302837226,
+      "eval_loss": 4.333541393280029,
+      "eval_runtime": 11.1986,
+      "eval_samples_per_second": 57.15,
+      "eval_steps_per_second": 9.555,
+      "eval_tts_loss": 5.138133705724894,
+      "step": 4000
+    },
+    {
+      "epoch": 0.08804347826086957,
+      "grad_norm": 2.684593677520752,
+      "learning_rate": 9.995640918676655e-05,
+      "loss": 4.6957,
+      "step": 4050
+    },
+    {
+      "epoch": 0.0891304347826087,
+      "grad_norm": 2.695699691772461,
+      "learning_rate": 9.995496561841154e-05,
+      "loss": 4.9025,
+      "step": 4100
+    },
+    {
+      "epoch": 0.09021739130434783,
+      "grad_norm": 2.6317203044891357,
+      "learning_rate": 9.99534985457503e-05,
+      "loss": 4.6769,
+      "step": 4150
+    },
+    {
+      "epoch": 0.09130434782608696,
+      "grad_norm": 2.2738523483276367,
+      "learning_rate": 9.995200796947313e-05,
+      "loss": 4.7356,
+      "step": 4200
+    },
+    {
+      "epoch": 0.09239130434782608,
+      "grad_norm": 3.4111077785491943,
+      "learning_rate": 9.995049389028132e-05,
+      "loss": 4.357,
+      "step": 4250
+    },
+    {
+      "epoch": 0.09347826086956522,
+      "grad_norm": 2.6470253467559814,
+      "learning_rate": 9.994895630888728e-05,
+      "loss": 4.3814,
+      "step": 4300
+    },
+    {
+      "epoch": 0.09456521739130434,
+      "grad_norm": 2.8316125869750977,
+      "learning_rate": 9.994739522601444e-05,
+      "loss": 4.3319,
+      "step": 4350
+    },
+    {
+      "epoch": 0.09565217391304348,
+      "grad_norm": 2.781243324279785,
+      "learning_rate": 9.994581064239732e-05,
+      "loss": 4.4887,
+      "step": 4400
+    },
+    {
+      "epoch": 0.0967391304347826,
+      "grad_norm": 4.024622917175293,
+      "learning_rate": 9.994420255878148e-05,
+      "loss": 4.4871,
+      "step": 4450
+    },
+    {
+      "epoch": 0.09782608695652174,
+      "grad_norm": 2.529017925262451,
+      "learning_rate": 9.994257097592352e-05,
+      "loss": 4.5259,
+      "step": 4500
+    },
+    {
+      "epoch": 0.09891304347826087,
+      "grad_norm": 1.8239754438400269,
+      "learning_rate": 9.994091589459116e-05,
+      "loss": 4.3442,
+      "step": 4550
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.8626890182495117,
+      "learning_rate": 9.993923731556308e-05,
+      "loss": 4.492,
+      "step": 4600
+    },
+    {
+      "epoch": 0.10108695652173913,
+      "grad_norm": 3.241543769836426,
+      "learning_rate": 9.99375352396291e-05,
+      "loss": 4.3023,
+      "step": 4650
+    },
+    {
+      "epoch": 0.10217391304347827,
+      "grad_norm": 3.9940199851989746,
+      "learning_rate": 9.993580966759007e-05,
+      "loss": 4.6912,
+      "step": 4700
+    },
+    {
+      "epoch": 0.10326086956521739,
+      "grad_norm": 2.9014837741851807,
+      "learning_rate": 9.993406060025787e-05,
+      "loss": 4.5711,
+      "step": 4750
+    },
+    {
+      "epoch": 0.10434782608695652,
+      "grad_norm": 1.6990643739700317,
+      "learning_rate": 9.993228803845545e-05,
+      "loss": 4.3312,
+      "step": 4800
+    },
+    {
+      "epoch": 0.10543478260869565,
+      "grad_norm": 2.805645704269409,
+      "learning_rate": 9.993049198301683e-05,
+      "loss": 4.4747,
+      "step": 4850
+    },
+    {
+      "epoch": 0.10652173913043478,
+      "grad_norm": 1.4846203327178955,
+      "learning_rate": 9.992867243478706e-05,
+      "loss": 4.4428,
+      "step": 4900
+    },
+    {
+      "epoch": 0.10760869565217392,
+      "grad_norm": 3.8630876541137695,
+      "learning_rate": 9.992682939462228e-05,
+      "loss": 4.8853,
+      "step": 4950
+    },
+    {
+      "epoch": 0.10869565217391304,
+      "grad_norm": 2.592350482940674,
+      "learning_rate": 9.992496286338963e-05,
+      "loss": 4.2777,
+      "step": 5000
+    },
+    {
+      "epoch": 0.10869565217391304,
+      "eval_asr_loss": 2.5512409255739876,
+      "eval_loss": 4.257580757141113,
+      "eval_runtime": 11.2845,
+      "eval_samples_per_second": 56.715,
+      "eval_steps_per_second": 9.482,
+      "eval_tts_loss": 5.1808239124319835,
+      "step": 5000
+    },
+    {
+      "epoch": 0.10978260869565218,
+      "grad_norm": 4.103634834289551,
+      "learning_rate": 9.992307284196735e-05,
+      "loss": 4.7146,
+      "step": 5050
+    },
+    {
+      "epoch": 0.1108695652173913,
+      "grad_norm": 2.421569585800171,
+      "learning_rate": 9.992115933124471e-05,
+      "loss": 4.6953,
+      "step": 5100
+    },
+    {
+      "epoch": 0.11195652173913044,
+      "grad_norm": 1.6327651739120483,
+      "learning_rate": 9.991922233212204e-05,
+      "loss": 4.4978,
+      "step": 5150
+    },
+    {
+      "epoch": 0.11304347826086956,
+      "grad_norm": 2.3584136962890625,
+      "learning_rate": 9.991726184551069e-05,
+      "loss": 4.3323,
+      "step": 5200
+    },
+    {
+      "epoch": 0.11413043478260869,
+      "grad_norm": 2.859476089477539,
+      "learning_rate": 9.991527787233313e-05,
+      "loss": 4.6582,
+      "step": 5250
+    },
+    {
+      "epoch": 0.11521739130434783,
+      "grad_norm": 2.6755621433258057,
+      "learning_rate": 9.991327041352278e-05,
+      "loss": 4.4851,
+      "step": 5300
+    },
+    {
+      "epoch": 0.11630434782608695,
+      "grad_norm": 2.0310680866241455,
+      "learning_rate": 9.991123947002425e-05,
+      "loss": 4.7106,
+      "step": 5350
+    },
+    {
+      "epoch": 0.11739130434782609,
+      "grad_norm": 1.8476403951644897,
+      "learning_rate": 9.990918504279307e-05,
+      "loss": 4.5227,
+      "step": 5400
+    },
+    {
+      "epoch": 0.11847826086956521,
+      "grad_norm": 1.8776793479919434,
+      "learning_rate": 9.990710713279586e-05,
+      "loss": 4.7484,
+      "step": 5450
+    },
+    {
+      "epoch": 0.11956521739130435,
+      "grad_norm": 2.704761266708374,
+      "learning_rate": 9.990500574101031e-05,
+      "loss": 4.7166,
+      "step": 5500
+    },
+    {
+      "epoch": 0.12065217391304348,
+      "grad_norm": 2.3094747066497803,
+      "learning_rate": 9.990288086842516e-05,
+      "loss": 4.2405,
+      "step": 5550
+    },
+    {
+      "epoch": 0.12173913043478261,
+      "grad_norm": 2.854323387145996,
+      "learning_rate": 9.990073251604018e-05,
+      "loss": 4.3237,
+      "step": 5600
+    },
+    {
+      "epoch": 0.12282608695652174,
+      "grad_norm": 1.8198304176330566,
+      "learning_rate": 9.989856068486615e-05,
+      "loss": 4.3826,
+      "step": 5650
+    },
+    {
+      "epoch": 0.12391304347826088,
+      "grad_norm": 2.3961496353149414,
+      "learning_rate": 9.989636537592501e-05,
+      "loss": 4.6476,
+      "step": 5700
+    },
+    {
+      "epoch": 0.125,
+      "grad_norm": 1.6423386335372925,
+      "learning_rate": 9.989414659024961e-05,
+      "loss": 4.1387,
+      "step": 5750
+    },
+    {
+      "epoch": 0.12608695652173912,
+      "grad_norm": 2.9040932655334473,
+      "learning_rate": 9.989190432888395e-05,
+      "loss": 4.2569,
+      "step": 5800
+    },
+    {
+      "epoch": 0.12717391304347825,
+      "grad_norm": 3.383302688598633,
+      "learning_rate": 9.9889638592883e-05,
+      "loss": 4.7406,
+      "step": 5850
+    },
+    {
+      "epoch": 0.1282608695652174,
+      "grad_norm": 5.475515842437744,
+      "learning_rate": 9.988734938331285e-05,
+      "loss": 4.2096,
+      "step": 5900
+    },
+    {
+      "epoch": 0.12934782608695652,
+      "grad_norm": 2.400712490081787,
+      "learning_rate": 9.988503670125058e-05,
+      "loss": 4.4473,
+      "step": 5950
+    },
+    {
+      "epoch": 0.13043478260869565,
+      "grad_norm": 2.487079381942749,
+      "learning_rate": 9.988270054778432e-05,
+      "loss": 4.5569,
+      "step": 6000
+    },
+    {
+      "epoch": 0.13043478260869565,
+      "eval_asr_loss": 2.6213980917950765,
+      "eval_loss": 4.200225830078125,
+      "eval_runtime": 11.1302,
+      "eval_samples_per_second": 57.501,
+      "eval_steps_per_second": 9.613,
+      "eval_tts_loss": 5.162653012560234,
+      "step": 6000
+    },
+    {
+      "epoch": 0.13152173913043477,
+      "grad_norm": 2.3784780502319336,
+      "learning_rate": 9.988034092401328e-05,
+      "loss": 4.3639,
+      "step": 6050
+    },
+    {
+      "epoch": 0.13260869565217392,
+      "grad_norm": 3.6094419956207275,
+      "learning_rate": 9.987795783104765e-05,
+      "loss": 4.2369,
+      "step": 6100
+    },
+    {
+      "epoch": 0.13369565217391305,
+      "grad_norm": 2.4041898250579834,
+      "learning_rate": 9.987555127000873e-05,
+      "loss": 4.4802,
+      "step": 6150
+    },
+    {
+      "epoch": 0.13478260869565217,
+      "grad_norm": 2.176236867904663,
+      "learning_rate": 9.987312124202881e-05,
+      "loss": 4.6597,
+      "step": 6200
+    },
+    {
+      "epoch": 0.1358695652173913,
+      "grad_norm": 1.969905972480774,
+      "learning_rate": 9.987066774825124e-05,
+      "loss": 4.4274,
+      "step": 6250
+    },
+    {
+      "epoch": 0.13695652173913042,
+      "grad_norm": 3.110654830932617,
+      "learning_rate": 9.986819078983045e-05,
+      "loss": 4.398,
+      "step": 6300
+    },
+    {
+      "epoch": 0.13804347826086957,
+      "grad_norm": 1.1360965967178345,
+      "learning_rate": 9.986569036793183e-05,
+      "loss": 4.2086,
+      "step": 6350
+    },
+    {
+      "epoch": 0.1391304347826087,
+      "grad_norm": 2.463719606399536,
+      "learning_rate": 9.986316648373187e-05,
+      "loss": 4.2981,
+      "step": 6400
+    },
+    {
+      "epoch": 0.14021739130434782,
+      "grad_norm": 2.486532211303711,
+      "learning_rate": 9.986061913841809e-05,
+      "loss": 4.2275,
+      "step": 6450
+    },
+    {
+      "epoch": 0.14130434782608695,
+      "grad_norm": 2.768862724304199,
+      "learning_rate": 9.985804833318903e-05,
+      "loss": 4.531,
+      "step": 6500
+    },
+    {
+      "epoch": 0.1423913043478261,
+      "grad_norm": 2.479682207107544,
+      "learning_rate": 9.985545406925428e-05,
+      "loss": 4.5276,
+      "step": 6550
+    },
+    {
+      "epoch": 0.14347826086956522,
+      "grad_norm": 2.016967296600342,
+      "learning_rate": 9.985283634783447e-05,
+      "loss": 4.3313,
+      "step": 6600
+    },
+    {
+      "epoch": 0.14456521739130435,
+      "grad_norm": 3.358776807785034,
+      "learning_rate": 9.985019517016127e-05,
+      "loss": 4.47,
+      "step": 6650
+    },
+    {
+      "epoch": 0.14565217391304347,
+      "grad_norm": 2.062159299850464,
+      "learning_rate": 9.984753053747738e-05,
+      "loss": 4.2664,
+      "step": 6700
+    },
+    {
+      "epoch": 0.14673913043478262,
+      "grad_norm": 2.3216474056243896,
+      "learning_rate": 9.984484245103652e-05,
+      "loss": 4.6873,
+      "step": 6750
+    },
+    {
+      "epoch": 0.14782608695652175,
+      "grad_norm": 3.587345600128174,
+      "learning_rate": 9.984213091210349e-05,
+      "loss": 4.2995,
+      "step": 6800
+    },
+    {
+      "epoch": 0.14891304347826087,
+      "grad_norm": 1.9871525764465332,
+      "learning_rate": 9.983939592195405e-05,
+      "loss": 4.4201,
+      "step": 6850
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 4.7601799964904785,
+      "learning_rate": 9.983663748187508e-05,
+      "loss": 4.2823,
+      "step": 6900
+    },
+    {
+      "epoch": 0.15108695652173912,
+      "grad_norm": 2.1823480129241943,
+      "learning_rate": 9.983385559316444e-05,
+      "loss": 4.5409,
+      "step": 6950
+    },
+    {
+      "epoch": 0.15217391304347827,
+      "grad_norm": 1.9459609985351562,
+      "learning_rate": 9.983105025713104e-05,
+      "loss": 4.3963,
+      "step": 7000
+    },
+    {
+      "epoch": 0.15217391304347827,
+      "eval_asr_loss": 2.64246700541829,
+      "eval_loss": 4.174106597900391,
+      "eval_runtime": 11.4091,
+      "eval_samples_per_second": 56.096,
+      "eval_steps_per_second": 9.378,
+      "eval_tts_loss": 5.271821806483866,
+      "step": 7000
+    },
+    {
+      "epoch": 0.1532608695652174,
+      "grad_norm": 3.8341915607452393,
+      "learning_rate": 9.982822147509481e-05,
+      "loss": 4.5849,
+      "step": 7050
+    },
+    {
+      "epoch": 0.15434782608695652,
+      "grad_norm": 2.2620415687561035,
+      "learning_rate": 9.982536924838672e-05,
+      "loss": 4.3853,
+      "step": 7100
+    },
+    {
+      "epoch": 0.15543478260869564,
+      "grad_norm": 3.4515111446380615,
+      "learning_rate": 9.98224935783488e-05,
+      "loss": 4.4753,
+      "step": 7150
+    },
+    {
+      "epoch": 0.1565217391304348,
+      "grad_norm": 2.8672986030578613,
+      "learning_rate": 9.981959446633404e-05,
+      "loss": 4.6994,
+      "step": 7200
+    },
+    {
+      "epoch": 0.15760869565217392,
+      "grad_norm": 2.919556140899658,
+      "learning_rate": 9.981667191370651e-05,
+      "loss": 4.5577,
+      "step": 7250
+    },
+    {
+      "epoch": 0.15869565217391304,
+      "grad_norm": 1.8371742963790894,
+      "learning_rate": 9.981372592184133e-05,
+      "loss": 4.0022,
+      "step": 7300
+    },
+    {
+      "epoch": 0.15978260869565217,
+      "grad_norm": 1.7818697690963745,
+      "learning_rate": 9.98107564921246e-05,
+      "loss": 4.3505,
+      "step": 7350
+    },
+    {
+      "epoch": 0.1608695652173913,
+      "grad_norm": 2.9086995124816895,
+      "learning_rate": 9.980776362595343e-05,
+      "loss": 4.0438,
+      "step": 7400
+    },
+    {
+      "epoch": 0.16195652173913044,
+      "grad_norm": 2.3632290363311768,
+      "learning_rate": 9.980474732473605e-05,
+      "loss": 4.1811,
+      "step": 7450
+    },
+    {
+      "epoch": 0.16304347826086957,
+      "grad_norm": 3.381775379180908,
+      "learning_rate": 9.980170758989164e-05,
+      "loss": 4.1882,
+      "step": 7500
+    },
+    {
+      "epoch": 0.1641304347826087,
+      "grad_norm": 2.8754525184631348,
+      "learning_rate": 9.97986444228504e-05,
+      "loss": 4.4118,
+      "step": 7550
+    },
+    {
+      "epoch": 0.16521739130434782,
+      "grad_norm": 1.8421484231948853,
+      "learning_rate": 9.979555782505363e-05,
+      "loss": 4.3014,
+      "step": 7600
+    },
+    {
+      "epoch": 0.16630434782608697,
+      "grad_norm": 2.170987844467163,
+      "learning_rate": 9.979244779795355e-05,
+      "loss": 4.5776,
+      "step": 7650
+    },
+    {
+      "epoch": 0.1673913043478261,
+      "grad_norm": 4.069926738739014,
+      "learning_rate": 9.978931434301351e-05,
+      "loss": 4.272,
+      "step": 7700
+    },
+    {
+      "epoch": 0.16847826086956522,
+      "grad_norm": 3.543511152267456,
+      "learning_rate": 9.97861574617078e-05,
+      "loss": 4.3141,
+      "step": 7750
+    },
+    {
+      "epoch": 0.16956521739130434,
+      "grad_norm": 1.1272289752960205,
+      "learning_rate": 9.978297715552177e-05,
+      "loss": 4.4087,
+      "step": 7800
+    },
+    {
+      "epoch": 0.17065217391304346,
+      "grad_norm": 1.961969256401062,
+      "learning_rate": 9.977977342595178e-05,
+      "loss": 4.4453,
+      "step": 7850
+    },
+    {
+      "epoch": 0.17173913043478262,
+      "grad_norm": 2.350886583328247,
+      "learning_rate": 9.97765462745052e-05,
+      "loss": 4.3917,
+      "step": 7900
+    },
+    {
+      "epoch": 0.17282608695652174,
+      "grad_norm": 1.1968051195144653,
+      "learning_rate": 9.977329570270051e-05,
+      "loss": 4.1804,
+      "step": 7950
+    },
+    {
+      "epoch": 0.17391304347826086,
+      "grad_norm": 1.2178881168365479,
+      "learning_rate": 9.977002171206706e-05,
+      "loss": 4.279,
+      "step": 8000
+    },
+    {
+      "epoch": 0.17391304347826086,
+      "eval_asr_loss": 2.6360731954683603,
+      "eval_loss": 4.113315582275391,
+      "eval_runtime": 11.2559,
+      "eval_samples_per_second": 56.859,
+      "eval_steps_per_second": 9.506,
+      "eval_tts_loss": 5.288750441627843,
+      "step": 8000
+    },
+    {
+      "epoch": 0.175,
+      "grad_norm": 2.2764134407043457,
+      "learning_rate": 9.976672430414531e-05,
+      "loss": 4.104,
+      "step": 8050
+    },
+    {
+      "epoch": 0.17608695652173914,
+      "grad_norm": 2.9129750728607178,
+      "learning_rate": 9.976340348048673e-05,
+      "loss": 4.3,
+      "step": 8100
+    },
+    {
+      "epoch": 0.17717391304347826,
+      "grad_norm": 2.2716591358184814,
+      "learning_rate": 9.976005924265384e-05,
+      "loss": 3.9896,
+      "step": 8150
+    },
+    {
+      "epoch": 0.1782608695652174,
+      "grad_norm": 2.305238962173462,
+      "learning_rate": 9.975669159222007e-05,
+      "loss": 4.4143,
+      "step": 8200
+    },
+    {
+      "epoch": 0.1793478260869565,
+      "grad_norm": 2.568418025970459,
+      "learning_rate": 9.975330053076997e-05,
+      "loss": 4.0646,
+      "step": 8250
+    },
+    {
+      "epoch": 0.18043478260869567,
+      "grad_norm": 3.0141923427581787,
+      "learning_rate": 9.974988605989907e-05,
+      "loss": 4.3971,
+      "step": 8300
+    },
+    {
+      "epoch": 0.1815217391304348,
+      "grad_norm": 2.3722753524780273,
+      "learning_rate": 9.97464481812139e-05,
+      "loss": 4.4071,
+      "step": 8350
+    },
+    {
+      "epoch": 0.1826086956521739,
+      "grad_norm": 1.8970969915390015,
+      "learning_rate": 9.974298689633201e-05,
+      "loss": 4.3679,
+      "step": 8400
+    },
+    {
+      "epoch": 0.18369565217391304,
+      "grad_norm": 1.5542092323303223,
+      "learning_rate": 9.9739502206882e-05,
+      "loss": 4.5544,
+      "step": 8450
+    },
+    {
+      "epoch": 0.18478260869565216,
+      "grad_norm": 3.2007203102111816,
+      "learning_rate": 9.973599411450342e-05,
+      "loss": 4.6362,
+      "step": 8500
+    },
+    {
+      "epoch": 0.1858695652173913,
+      "grad_norm": 2.7745096683502197,
+      "learning_rate": 9.973246262084688e-05,
+      "loss": 4.3841,
+      "step": 8550
+    },
+    {
+      "epoch": 0.18695652173913044,
+      "grad_norm": 2.288344383239746,
+      "learning_rate": 9.972890772757397e-05,
+      "loss": 4.3282,
+      "step": 8600
+    },
+    {
+      "epoch": 0.18804347826086956,
+      "grad_norm": 1.7431132793426514,
+      "learning_rate": 9.97253294363573e-05,
+      "loss": 4.6419,
+      "step": 8650
+    },
+    {
+      "epoch": 0.1891304347826087,
+      "grad_norm": 2.577338933944702,
+      "learning_rate": 9.97217277488805e-05,
+      "loss": 4.4464,
+      "step": 8700
+    },
+    {
+      "epoch": 0.19021739130434784,
+      "grad_norm": 1.9421231746673584,
+      "learning_rate": 9.97181026668382e-05,
+      "loss": 4.428,
+      "step": 8750
+    },
+    {
+      "epoch": 0.19130434782608696,
+      "grad_norm": 2.9982190132141113,
+      "learning_rate": 9.971445419193603e-05,
+      "loss": 4.6251,
+      "step": 8800
+    },
+    {
+      "epoch": 0.1923913043478261,
+      "grad_norm": 4.028087615966797,
+      "learning_rate": 9.971078232589067e-05,
+      "loss": 4.3337,
+      "step": 8850
+    },
+    {
+      "epoch": 0.1934782608695652,
+      "grad_norm": 2.3939244747161865,
+      "learning_rate": 9.970708707042971e-05,
+      "loss": 4.2349,
+      "step": 8900
+    },
+    {
+      "epoch": 0.19456521739130433,
+      "grad_norm": 1.9768601655960083,
+      "learning_rate": 9.970336842729184e-05,
+      "loss": 4.3149,
+      "step": 8950
+    },
+    {
+      "epoch": 0.1956521739130435,
+      "grad_norm": 3.1762466430664062,
+      "learning_rate": 9.969962639822671e-05,
+      "loss": 4.4202,
+      "step": 9000
+    },
+    {
+      "epoch": 0.1956521739130435,
+      "eval_asr_loss": 2.629300301255897,
+      "eval_loss": 4.0937113761901855,
+      "eval_runtime": 11.792,
+      "eval_samples_per_second": 54.274,
+      "eval_steps_per_second": 9.074,
+      "eval_tts_loss": 5.314797203742016,
+      "step": 9000
+    },
+    {
+      "epoch": 0.1967391304347826,
+      "grad_norm": 1.8454028367996216,
+      "learning_rate": 9.9695860984995e-05,
+      "loss": 4.139,
+      "step": 9050
+    },
+    {
+      "epoch": 0.19782608695652174,
+      "grad_norm": 2.5526962280273438,
+      "learning_rate": 9.969207218936834e-05,
+      "loss": 4.3805,
+      "step": 9100
+    },
+    {
+      "epoch": 0.19891304347826086,
+      "grad_norm": 1.42033851146698,
+      "learning_rate": 9.96882600131294e-05,
+      "loss": 4.28,
+      "step": 9150
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.136204957962036,
+      "learning_rate": 9.968442445807188e-05,
+      "loss": 4.3775,
+      "step": 9200
+    },
+    {
+      "epoch": 0.20108695652173914,
+      "grad_norm": 1.782203197479248,
+      "learning_rate": 9.968056552600043e-05,
+      "loss": 4.2135,
+      "step": 9250
+    },
+    {
+      "epoch": 0.20217391304347826,
+      "grad_norm": 3.0619659423828125,
+      "learning_rate": 9.96766832187307e-05,
+      "loss": 4.4359,
+      "step": 9300
+    },
+    {
+      "epoch": 0.20326086956521738,
+      "grad_norm": 3.9900858402252197,
+      "learning_rate": 9.967277753808936e-05,
+      "loss": 4.0223,
+      "step": 9350
+    },
+    {
+      "epoch": 0.20434782608695654,
+      "grad_norm": 2.068382501602173,
+      "learning_rate": 9.96688484859141e-05,
+      "loss": 3.9307,
+      "step": 9400
+    },
+    {
+      "epoch": 0.20543478260869566,
+      "grad_norm": 2.419823408126831,
+      "learning_rate": 9.966489606405355e-05,
+      "loss": 4.2665,
+      "step": 9450
+    },
+    {
+      "epoch": 0.20652173913043478,
+      "grad_norm": 2.9332566261291504,
+      "learning_rate": 9.966092027436736e-05,
+      "loss": 4.2707,
+      "step": 9500
+    },
+    {
+      "epoch": 0.2076086956521739,
+      "grad_norm": 2.023785352706909,
+      "learning_rate": 9.96569211187262e-05,
+      "loss": 4.1401,
+      "step": 9550
+    },
+    {
+      "epoch": 0.20869565217391303,
+      "grad_norm": 2.350621461868286,
+      "learning_rate": 9.965289859901169e-05,
+      "loss": 4.2363,
+      "step": 9600
+    },
+    {
+      "epoch": 0.20978260869565218,
+      "grad_norm": 3.8558528423309326,
+      "learning_rate": 9.964885271711649e-05,
+      "loss": 3.8054,
+      "step": 9650
+    },
+    {
+      "epoch": 0.2108695652173913,
+      "grad_norm": 1.7765642404556274,
+      "learning_rate": 9.964478347494419e-05,
+      "loss": 4.3695,
+      "step": 9700
+    },
+    {
+      "epoch": 0.21195652173913043,
+      "grad_norm": 2.4970788955688477,
+      "learning_rate": 9.964069087440945e-05,
+      "loss": 4.2529,
+      "step": 9750
+    },
+    {
+      "epoch": 0.21304347826086956,
+      "grad_norm": 2.3086397647857666,
+      "learning_rate": 9.963657491743785e-05,
+      "loss": 4.2548,
+      "step": 9800
+    },
+    {
+      "epoch": 0.2141304347826087,
+      "grad_norm": 1.9614365100860596,
+      "learning_rate": 9.963243560596601e-05,
+      "loss": 4.2051,
+      "step": 9850
+    },
+    {
+      "epoch": 0.21521739130434783,
+      "grad_norm": 2.378877639770508,
+      "learning_rate": 9.96282729419415e-05,
+      "loss": 4.1855,
+      "step": 9900
+    },
+    {
+      "epoch": 0.21630434782608696,
+      "grad_norm": 1.7504818439483643,
+      "learning_rate": 9.96240869273229e-05,
+      "loss": 4.2168,
+      "step": 9950
+    },
+    {
+      "epoch": 0.21739130434782608,
+      "grad_norm": 2.1233267784118652,
+      "learning_rate": 9.961987756407978e-05,
+      "loss": 4.4443,
+      "step": 10000
+    },
+    {
+      "epoch": 0.21739130434782608,
+      "eval_asr_loss": 2.591305898835824,
+      "eval_loss": 4.0733113288879395,
+      "eval_runtime": 11.2808,
+      "eval_samples_per_second": 56.733,
+      "eval_steps_per_second": 9.485,
+      "eval_tts_loss": 5.32374426206258,
+      "step": 10000
+    },
+    {
+      "epoch": 0.2184782608695652,
+      "grad_norm": 2.516474723815918,
+      "learning_rate": 9.961564485419268e-05,
+      "loss": 4.298,
+      "step": 10050
+    },
+    {
+      "epoch": 0.21956521739130436,
+      "grad_norm": 1.9682726860046387,
+      "learning_rate": 9.961138879965312e-05,
+      "loss": 4.2673,
+      "step": 10100
+    },
+    {
+      "epoch": 0.22065217391304348,
+      "grad_norm": 3.1358115673065186,
+      "learning_rate": 9.960710940246363e-05,
+      "loss": 3.9091,
+      "step": 10150
+    },
+    {
+      "epoch": 0.2217391304347826,
+      "grad_norm": 3.4082937240600586,
+      "learning_rate": 9.960280666463773e-05,
+      "loss": 4.3823,
+      "step": 10200
+    },
+    {
+      "epoch": 0.22282608695652173,
+      "grad_norm": 2.249570369720459,
+      "learning_rate": 9.959848058819986e-05,
+      "loss": 4.1687,
+      "step": 10250
+    },
+    {
+      "epoch": 0.22391304347826088,
+      "grad_norm": 1.4500969648361206,
+      "learning_rate": 9.95941311751855e-05,
+      "loss": 4.4216,
+      "step": 10300
+    },
+    {
+      "epoch": 0.225,
+      "grad_norm": 1.9060927629470825,
+      "learning_rate": 9.958975842764108e-05,
+      "loss": 4.2522,
+      "step": 10350
+    },
+    {
+      "epoch": 0.22608695652173913,
+      "grad_norm": 1.9281275272369385,
+      "learning_rate": 9.958536234762404e-05,
+      "loss": 4.2719,
+      "step": 10400
+    },
+    {
+      "epoch": 0.22717391304347825,
+      "grad_norm": 2.0862088203430176,
+      "learning_rate": 9.958094293720277e-05,
+      "loss": 4.0065,
+      "step": 10450
+    },
+    {
+      "epoch": 0.22826086956521738,
+      "grad_norm": 1.934340000152588,
+      "learning_rate": 9.957650019845665e-05,
+      "loss": 4.2551,
+      "step": 10500
+    },
+    {
+      "epoch": 0.22934782608695653,
+      "grad_norm": 3.107830047607422,
+      "learning_rate": 9.957203413347602e-05,
+      "loss": 4.2608,
+      "step": 10550
+    },
+    {
+      "epoch": 0.23043478260869565,
+      "grad_norm": 3.3998148441314697,
+      "learning_rate": 9.956754474436223e-05,
+      "loss": 4.1243,
+      "step": 10600
+    },
+    {
+      "epoch": 0.23152173913043478,
+      "grad_norm": 1.9935047626495361,
+      "learning_rate": 9.956303203322756e-05,
+      "loss": 4.55,
+      "step": 10650
+    },
+    {
+      "epoch": 0.2326086956521739,
+      "grad_norm": 2.274759531021118,
+      "learning_rate": 9.955849600219531e-05,
+      "loss": 4.5959,
+      "step": 10700
+    },
+    {
+      "epoch": 0.23369565217391305,
+      "grad_norm": 2.2281334400177,
+      "learning_rate": 9.955393665339971e-05,
+      "loss": 4.3989,
+      "step": 10750
+    },
+    {
+      "epoch": 0.23478260869565218,
+      "grad_norm": 1.642350435256958,
+      "learning_rate": 9.954935398898597e-05,
+      "loss": 4.4348,
+      "step": 10800
+    },
+    {
+      "epoch": 0.2358695652173913,
+      "grad_norm": 1.8325356245040894,
+      "learning_rate": 9.954474801111031e-05,
+      "loss": 4.1066,
+      "step": 10850
+    },
+    {
+      "epoch": 0.23695652173913043,
+      "grad_norm": 1.2226781845092773,
+      "learning_rate": 9.954011872193984e-05,
+      "loss": 4.1221,
+      "step": 10900
+    },
+    {
+      "epoch": 0.23804347826086958,
+      "grad_norm": 2.1517810821533203,
+      "learning_rate": 9.953546612365274e-05,
+      "loss": 3.9762,
+      "step": 10950
+    },
+    {
+      "epoch": 0.2391304347826087,
+      "grad_norm": 2.18406081199646,
+      "learning_rate": 9.953079021843808e-05,
+      "loss": 4.2836,
+      "step": 11000
+    },
+    {
+      "epoch": 0.2391304347826087,
+      "eval_asr_loss": 2.598486068728943,
+      "eval_loss": 4.054217338562012,
+      "eval_runtime": 11.358,
+      "eval_samples_per_second": 56.348,
+      "eval_steps_per_second": 9.421,
+      "eval_tts_loss": 5.356123704782253,
+      "step": 11000
+    },
+    {
+      "epoch": 0.24021739130434783,
+      "grad_norm": 2.1122887134552,
+      "learning_rate": 9.95260910084959e-05,
+      "loss": 4.0984,
+      "step": 11050
+    },
+    {
+      "epoch": 0.24130434782608695,
+      "grad_norm": 1.9920451641082764,
+      "learning_rate": 9.952136849603726e-05,
+      "loss": 4.3487,
+      "step": 11100
+    },
+    {
+      "epoch": 0.24239130434782608,
+      "grad_norm": 3.9625918865203857,
+      "learning_rate": 9.951662268328413e-05,
+      "loss": 4.1589,
+      "step": 11150
+    },
+    {
+      "epoch": 0.24347826086956523,
+      "grad_norm": 1.6680744886398315,
+      "learning_rate": 9.951185357246944e-05,
+      "loss": 4.1001,
+      "step": 11200
+    },
+    {
+      "epoch": 0.24456521739130435,
+      "grad_norm": 1.888364553451538,
+      "learning_rate": 9.950706116583715e-05,
+      "loss": 4.4796,
+      "step": 11250
+    },
+    {
+      "epoch": 0.24565217391304348,
+      "grad_norm": 1.6696151494979858,
+      "learning_rate": 9.950224546564211e-05,
+      "loss": 4.1396,
+      "step": 11300
+    },
+    {
+      "epoch": 0.2467391304347826,
+      "grad_norm": 1.895444631576538,
+      "learning_rate": 9.949740647415016e-05,
+      "loss": 4.3311,
+      "step": 11350
+    },
+    {
+      "epoch": 0.24782608695652175,
+      "grad_norm": 2.224364995956421,
+      "learning_rate": 9.949254419363808e-05,
+      "loss": 4.3021,
+      "step": 11400
+    },
+    {
+      "epoch": 0.24891304347826088,
+      "grad_norm": 2.5382299423217773,
+      "learning_rate": 9.948765862639365e-05,
+      "loss": 4.113,
+      "step": 11450
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.1696720123291016,
+      "learning_rate": 9.948274977471555e-05,
+      "loss": 4.3802,
+      "step": 11500
+    },
+    {
+      "epoch": 0.2510869565217391,
+      "grad_norm": 2.0751821994781494,
+      "learning_rate": 9.947781764091344e-05,
+      "loss": 4.3058,
+      "step": 11550
+    },
+    {
+      "epoch": 0.25217391304347825,
+      "grad_norm": 2.339843273162842,
+      "learning_rate": 9.947286222730797e-05,
+      "loss": 4.2898,
+      "step": 11600
+    },
+    {
+      "epoch": 0.2532608695652174,
+      "grad_norm": 2.529846668243408,
+      "learning_rate": 9.946788353623068e-05,
+      "loss": 4.224,
+      "step": 11650
+    },
+    {
+      "epoch": 0.2543478260869565,
+      "grad_norm": 2.0904839038848877,
+      "learning_rate": 9.946288157002412e-05,
+      "loss": 3.9864,
+      "step": 11700
+    },
+    {
+      "epoch": 0.2554347826086957,
+      "grad_norm": 1.972814917564392,
+      "learning_rate": 9.945785633104174e-05,
+      "loss": 4.1864,
+      "step": 11750
+    },
+    {
+      "epoch": 0.2565217391304348,
+      "grad_norm": 2.7812299728393555,
+      "learning_rate": 9.945280782164798e-05,
+      "loss": 4.4875,
+      "step": 11800
+    },
+    {
+      "epoch": 0.2576086956521739,
+      "grad_norm": 2.132671594619751,
+      "learning_rate": 9.94477360442182e-05,
+      "loss": 4.4964,
+      "step": 11850
+    },
+    {
+      "epoch": 0.25869565217391305,
+      "grad_norm": 2.1867575645446777,
+      "learning_rate": 9.944264100113874e-05,
+      "loss": 4.2672,
+      "step": 11900
+    },
+    {
+      "epoch": 0.2597826086956522,
+      "grad_norm": 2.330216407775879,
+      "learning_rate": 9.943752269480686e-05,
+      "loss": 4.3042,
+      "step": 11950
+    },
+    {
+      "epoch": 0.2608695652173913,
+      "grad_norm": 2.111233949661255,
+      "learning_rate": 9.943238112763076e-05,
+      "loss": 4.1073,
+      "step": 12000
+    },
+    {
+      "epoch": 0.2608695652173913,
+      "eval_asr_loss": 2.575289285385664,
+      "eval_loss": 4.001055717468262,
+      "eval_runtime": 11.6371,
+      "eval_samples_per_second": 54.996,
+      "eval_steps_per_second": 9.195,
+      "eval_tts_loss": 5.413907140913973,
+      "step": 12000
+    },
+    {
+      "epoch": 0.2619565217391304,
+      "grad_norm": 3.3587117195129395,
+      "learning_rate": 9.942721630202963e-05,
+      "loss": 4.1906,
+      "step": 12050
+    },
+    {
+      "epoch": 0.26304347826086955,
+      "grad_norm": 2.4953081607818604,
+      "learning_rate": 9.942202822043354e-05,
+      "loss": 3.8885,
+      "step": 12100
+    },
+    {
+      "epoch": 0.26413043478260867,
+      "grad_norm": 4.285083293914795,
+      "learning_rate": 9.941681688528354e-05,
+      "loss": 4.5575,
+      "step": 12150
+    },
+    {
+      "epoch": 0.26521739130434785,
+      "grad_norm": 1.7230956554412842,
+      "learning_rate": 9.941158229903163e-05,
+      "loss": 4.2665,
+      "step": 12200
+    },
+    {
+      "epoch": 0.266304347826087,
+      "grad_norm": 1.0887356996536255,
+      "learning_rate": 9.940632446414073e-05,
+      "loss": 4.4211,
+      "step": 12250
+    },
+    {
+      "epoch": 0.2673913043478261,
+      "grad_norm": 2.4916415214538574,
+      "learning_rate": 9.940104338308467e-05,
+      "loss": 4.2423,
+      "step": 12300
+    },
+    {
+      "epoch": 0.2684782608695652,
+      "grad_norm": 4.6231770515441895,
+      "learning_rate": 9.939573905834831e-05,
+      "loss": 3.8214,
+      "step": 12350
+    },
+    {
+      "epoch": 0.26956521739130435,
+      "grad_norm": 2.0836164951324463,
+      "learning_rate": 9.939041149242735e-05,
+      "loss": 4.1328,
+      "step": 12400
+    },
+    {
+      "epoch": 0.27065217391304347,
+      "grad_norm": 3.0433685779571533,
+      "learning_rate": 9.938506068782847e-05,
+      "loss": 3.911,
+      "step": 12450
+    },
+    {
+      "epoch": 0.2717391304347826,
+      "grad_norm": 1.9049893617630005,
+      "learning_rate": 9.937968664706927e-05,
+      "loss": 4.116,
+      "step": 12500
+    },
+    {
+      "epoch": 0.2728260869565217,
+      "grad_norm": 1.5418988466262817,
+      "learning_rate": 9.937428937267831e-05,
+      "loss": 4.6051,
+      "step": 12550
+    },
+    {
+      "epoch": 0.27391304347826084,
+      "grad_norm": 2.2141311168670654,
+      "learning_rate": 9.936886886719503e-05,
+      "loss": 3.9949,
+      "step": 12600
+    },
+    {
+      "epoch": 0.275,
+      "grad_norm": 1.5093140602111816,
+      "learning_rate": 9.936342513316987e-05,
+      "loss": 4.2754,
+      "step": 12650
+    },
+    {
+      "epoch": 0.27608695652173915,
+      "grad_norm": 3.3034720420837402,
+      "learning_rate": 9.935795817316411e-05,
+      "loss": 4.5361,
+      "step": 12700
+    },
+    {
+      "epoch": 0.27717391304347827,
+      "grad_norm": 1.627867341041565,
+      "learning_rate": 9.935246798975007e-05,
+      "loss": 4.0012,
+      "step": 12750
+    },
+    {
+      "epoch": 0.2782608695652174,
+      "grad_norm": 2.3469769954681396,
+      "learning_rate": 9.934695458551089e-05,
+      "loss": 3.7422,
+      "step": 12800
+    },
+    {
+      "epoch": 0.2793478260869565,
+      "grad_norm": 2.5750248432159424,
+      "learning_rate": 9.934141796304068e-05,
+      "loss": 4.1241,
+      "step": 12850
+    },
+    {
+      "epoch": 0.28043478260869564,
+      "grad_norm": 1.8672727346420288,
+      "learning_rate": 9.93358581249445e-05,
+      "loss": 4.445,
+      "step": 12900
+    },
+    {
+      "epoch": 0.28152173913043477,
+      "grad_norm": 1.281456470489502,
+      "learning_rate": 9.933027507383831e-05,
+      "loss": 4.1293,
+      "step": 12950
+    },
+    {
+      "epoch": 0.2826086956521739,
+      "grad_norm": 1.8554446697235107,
+      "learning_rate": 9.932466881234898e-05,
+      "loss": 4.0762,
+      "step": 13000
+    },
+    {
+      "epoch": 0.2826086956521739,
+      "eval_asr_loss": 2.603479289659652,
+      "eval_loss": 3.983154296875,
+      "eval_runtime": 11.3213,
+      "eval_samples_per_second": 56.531,
+      "eval_steps_per_second": 9.451,
+      "eval_tts_loss": 5.388899450382259,
+      "step": 13000
+    },
+    {
+      "epoch": 0.28369565217391307,
+      "grad_norm": 2.137413263320923,
+      "learning_rate": 9.931903934311431e-05,
+      "loss": 3.9685,
+      "step": 13050
+    },
+    {
+      "epoch": 0.2847826086956522,
+      "grad_norm": 2.4471590518951416,
+      "learning_rate": 9.9313386668783e-05,
+      "loss": 4.4134,
+      "step": 13100
+    },
+    {
+      "epoch": 0.2858695652173913,
+      "grad_norm": 1.5745162963867188,
+      "learning_rate": 9.930771079201474e-05,
+      "loss": 4.036,
+      "step": 13150
+    },
+    {
+      "epoch": 0.28695652173913044,
+      "grad_norm": 3.4924018383026123,
+      "learning_rate": 9.930201171548006e-05,
+      "loss": 4.3891,
+      "step": 13200
+    },
+    {
+      "epoch": 0.28804347826086957,
+      "grad_norm": 3.474689245223999,
+      "learning_rate": 9.929628944186042e-05,
+      "loss": 4.2206,
+      "step": 13250
+    },
+    {
+      "epoch": 0.2891304347826087,
+      "grad_norm": 2.069572687149048,
+      "learning_rate": 9.929054397384822e-05,
+      "loss": 4.3427,
+      "step": 13300
+    },
+    {
+      "epoch": 0.2902173913043478,
+      "grad_norm": 3.209120988845825,
+      "learning_rate": 9.928477531414675e-05,
+      "loss": 4.3438,
+      "step": 13350
+    },
+    {
+      "epoch": 0.29130434782608694,
+      "grad_norm": 2.3992714881896973,
+      "learning_rate": 9.927898346547023e-05,
+      "loss": 3.9973,
+      "step": 13400
+    },
+    {
+      "epoch": 0.29239130434782606,
+      "grad_norm": 4.438747882843018,
+      "learning_rate": 9.927316843054379e-05,
+      "loss": 3.7041,
+      "step": 13450
+    },
+    {
+      "epoch": 0.29347826086956524,
+      "grad_norm": 2.1427340507507324,
+      "learning_rate": 9.926733021210344e-05,
+      "loss": 4.1536,
+      "step": 13500
+    },
+    {
+      "epoch": 0.29456521739130437,
+      "grad_norm": 2.3441412448883057,
+      "learning_rate": 9.926146881289613e-05,
+      "loss": 3.9823,
+      "step": 13550
+    },
+    {
+      "epoch": 0.2956521739130435,
+      "grad_norm": 1.3850131034851074,
+      "learning_rate": 9.925558423567973e-05,
+      "loss": 4.0567,
+      "step": 13600
+    },
+    {
+      "epoch": 0.2967391304347826,
+      "grad_norm": 1.8057410717010498,
+      "learning_rate": 9.924967648322294e-05,
+      "loss": 4.2215,
+      "step": 13650
+    },
+    {
+      "epoch": 0.29782608695652174,
+      "grad_norm": 2.371821880340576,
+      "learning_rate": 9.924374555830547e-05,
+      "loss": 4.4935,
+      "step": 13700
+    },
+    {
+      "epoch": 0.29891304347826086,
+      "grad_norm": 2.152357339859009,
+      "learning_rate": 9.923779146371786e-05,
+      "loss": 4.3294,
+      "step": 13750
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.7699871063232422,
+      "learning_rate": 9.923181420226155e-05,
+      "loss": 4.2102,
+      "step": 13800
+    },
+    {
+      "epoch": 0.3010869565217391,
+      "grad_norm": 1.396740436553955,
+      "learning_rate": 9.922581377674891e-05,
+      "loss": 3.8649,
+      "step": 13850
+    },
+    {
+      "epoch": 0.30217391304347824,
+      "grad_norm": 1.686732530593872,
+      "learning_rate": 9.921991088869989e-05,
+      "loss": 4.2365,
+      "step": 13900
+    },
+    {
+      "epoch": 0.3032608695652174,
+      "grad_norm": 2.4129815101623535,
+      "learning_rate": 9.921386460669542e-05,
+      "loss": 4.1552,
+      "step": 13950
+    },
+    {
+      "epoch": 0.30434782608695654,
+      "grad_norm": 1.9270827770233154,
+      "learning_rate": 9.92077951690801e-05,
+      "loss": 4.3371,
+      "step": 14000
+    },
+    {
+      "epoch": 0.30434782608695654,
+      "eval_asr_loss": 2.621762472140386,
+      "eval_loss": 3.973423957824707,
+      "eval_runtime": 11.5586,
+      "eval_samples_per_second": 55.37,
+      "eval_steps_per_second": 9.257,
+      "eval_tts_loss": 5.440839271461044,
+      "step": 14000
+    },
+    {
+      "epoch": 0.30543478260869567,
+      "grad_norm": 2.034511089324951,
+      "learning_rate": 9.920170257870962e-05,
+      "loss": 4.0571,
+      "step": 14050
+    },
+    {
+      "epoch": 0.3065217391304348,
+      "grad_norm": 1.5555872917175293,
+      "learning_rate": 9.919558683845066e-05,
+      "loss": 4.2282,
+      "step": 14100
+    },
+    {
+      "epoch": 0.3076086956521739,
+      "grad_norm": 1.7540497779846191,
+      "learning_rate": 9.918944795118071e-05,
+      "loss": 4.2599,
+      "step": 14150
+    },
+    {
+      "epoch": 0.30869565217391304,
+      "grad_norm": 1.7837488651275635,
+      "learning_rate": 9.918328591978816e-05,
+      "loss": 4.0573,
+      "step": 14200
+    },
+    {
+      "epoch": 0.30978260869565216,
+      "grad_norm": 1.2670475244522095,
+      "learning_rate": 9.917710074717233e-05,
+      "loss": 4.0929,
+      "step": 14250
+    },
+    {
+      "epoch": 0.3108695652173913,
+      "grad_norm": 1.9616267681121826,
+      "learning_rate": 9.91708924362434e-05,
+      "loss": 4.0503,
+      "step": 14300
+    },
+    {
+      "epoch": 0.3119565217391304,
+      "grad_norm": 2.7847957611083984,
+      "learning_rate": 9.916466098992242e-05,
+      "loss": 3.9891,
+      "step": 14350
+    },
+    {
+      "epoch": 0.3130434782608696,
+      "grad_norm": 2.8416688442230225,
+      "learning_rate": 9.91584064111414e-05,
+      "loss": 4.0328,
+      "step": 14400
+    },
+    {
+      "epoch": 0.3141304347826087,
+      "grad_norm": 1.727723479270935,
+      "learning_rate": 9.915212870284314e-05,
+      "loss": 3.9954,
+      "step": 14450
+    },
+    {
+      "epoch": 0.31521739130434784,
+      "grad_norm": 2.170253276824951,
+      "learning_rate": 9.914582786798136e-05,
+      "loss": 4.3862,
+      "step": 14500
+    },
+    {
+      "epoch": 0.31630434782608696,
+      "grad_norm": 1.7831149101257324,
+      "learning_rate": 9.91395039095207e-05,
+      "loss": 4.2361,
+      "step": 14550
+    },
+    {
+      "epoch": 0.3173913043478261,
+      "grad_norm": 1.8148229122161865,
+      "learning_rate": 9.913315683043662e-05,
+      "loss": 4.0333,
+      "step": 14600
+    },
+    {
+      "epoch": 0.3184782608695652,
+      "grad_norm": 1.904608964920044,
+      "learning_rate": 9.912678663371547e-05,
+      "loss": 4.4022,
+      "step": 14650
+    },
+    {
+      "epoch": 0.31956521739130433,
+      "grad_norm": 3.0111865997314453,
+      "learning_rate": 9.912039332235452e-05,
+      "loss": 4.0977,
+      "step": 14700
+    },
+    {
+      "epoch": 0.32065217391304346,
+      "grad_norm": 2.758352756500244,
+      "learning_rate": 9.911397689936186e-05,
+      "loss": 4.1508,
+      "step": 14750
+    },
+    {
+      "epoch": 0.3217391304347826,
+      "grad_norm": 1.5165542364120483,
+      "learning_rate": 9.91075373677565e-05,
+      "loss": 4.0186,
+      "step": 14800
+    },
+    {
+      "epoch": 0.32282608695652176,
+      "grad_norm": 2.3647053241729736,
+      "learning_rate": 9.910107473056832e-05,
+      "loss": 4.324,
+      "step": 14850
+    },
+    {
+      "epoch": 0.3239130434782609,
+      "grad_norm": 2.138179302215576,
+      "learning_rate": 9.9094588990838e-05,
+      "loss": 4.4301,
+      "step": 14900
+    },
+    {
+      "epoch": 0.325,
+      "grad_norm": 2.526128053665161,
+      "learning_rate": 9.90880801516172e-05,
+      "loss": 4.4203,
+      "step": 14950
+    },
+    {
+      "epoch": 0.32608695652173914,
+      "grad_norm": 2.138474464416504,
+      "learning_rate": 9.908154821596837e-05,
+      "loss": 4.2592,
+      "step": 15000
+    },
+    {
+      "epoch": 0.32608695652173914,
+      "eval_asr_loss": 2.6088158428552903,
+      "eval_loss": 3.9580538272857666,
+      "eval_runtime": 11.3769,
+      "eval_samples_per_second": 56.255,
+      "eval_steps_per_second": 9.405,
+      "eval_tts_loss": 5.412408770328585,
+      "step": 15000
+    },
+    {
+      "epoch": 0.32717391304347826,
+      "grad_norm": 3.239708423614502,
+      "learning_rate": 9.907499318696485e-05,
+      "loss": 3.8169,
+      "step": 15050
+    },
+    {
+      "epoch": 0.3282608695652174,
+      "grad_norm": 1.7221394777297974,
+      "learning_rate": 9.906841506769084e-05,
+      "loss": 4.0436,
+      "step": 15100
+    },
+    {
+      "epoch": 0.3293478260869565,
+      "grad_norm": 1.565096378326416,
+      "learning_rate": 9.906181386124141e-05,
+      "loss": 4.3082,
+      "step": 15150
+    },
+    {
+      "epoch": 0.33043478260869563,
+      "grad_norm": 1.188761591911316,
+      "learning_rate": 9.905518957072252e-05,
+      "loss": 3.938,
+      "step": 15200
+    },
+    {
+      "epoch": 0.33152173913043476,
+      "grad_norm": 2.034902811050415,
+      "learning_rate": 9.904854219925093e-05,
+      "loss": 4.0912,
+      "step": 15250
+    },
+    {
+      "epoch": 0.33260869565217394,
+      "grad_norm": 2.4012644290924072,
+      "learning_rate": 9.904187174995432e-05,
+      "loss": 4.3901,
+      "step": 15300
+    },
+    {
+      "epoch": 0.33369565217391306,
+      "grad_norm": 1.7025322914123535,
+      "learning_rate": 9.903517822597119e-05,
+      "loss": 4.1072,
+      "step": 15350
+    },
+    {
+      "epoch": 0.3347826086956522,
+      "grad_norm": 2.4525983333587646,
+      "learning_rate": 9.90284616304509e-05,
+      "loss": 4.5251,
+      "step": 15400
+    },
+    {
+      "epoch": 0.3358695652173913,
+      "grad_norm": 1.9360466003417969,
+      "learning_rate": 9.90217219665537e-05,
+      "loss": 4.1684,
+      "step": 15450
+    },
+    {
+      "epoch": 0.33695652173913043,
+      "grad_norm": 2.3561511039733887,
+      "learning_rate": 9.901495923745066e-05,
+      "loss": 3.9902,
+      "step": 15500
+    }
+  ],
+  "logging_steps": 50,
+  "max_steps": 230000,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 5,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 2.8765353560296653e+17,
+  "train_batch_size": 6,
+  "trial_name": null,
+  "trial_params": null
+}