Upload kaggle_output.json with huggingface_hub
Browse files- kaggle_output.json +5 -5
kaggle_output.json
CHANGED
|
@@ -4,7 +4,7 @@
|
|
| 4 |
"base_model": "Qwen/Qwen2.5-3B-Instruct",
|
| 5 |
"adapter_path": "/kaggle/working/slow_consolidation_train/adapter",
|
| 6 |
"output_json": "/kaggle/working/slow_consolidation_train/kaggle_output.json",
|
| 7 |
-
"loss":
|
| 8 |
"sft_source": "local",
|
| 9 |
"source_note": "M1 consolidation_sft.jsonl (lokal, §10.3)",
|
| 10 |
"dataset": "consolidation_sft.jsonl",
|
|
@@ -21,15 +21,15 @@
|
|
| 21 |
"num_train_epochs": 15.0,
|
| 22 |
"max_length": 1024,
|
| 23 |
"learning_rate": 0.0002,
|
| 24 |
-
"lora_r":
|
| 25 |
-
"lora_alpha":
|
| 26 |
"compute_dtype": "float16",
|
| 27 |
"amp": "disabled (fp16=False, bf16=False)",
|
| 28 |
"compute_capability": "sm_75",
|
| 29 |
"model_init_dtype": "float16",
|
| 30 |
"gpu": "Tesla T4",
|
| 31 |
-
"started_utc": "2026-10-
|
| 32 |
-
"finished_utc": "2026-10-
|
| 33 |
"library_versions": {
|
| 34 |
"transformers": "5.16.1",
|
| 35 |
"peft": "0.20.0",
|
|
|
|
| 4 |
"base_model": "Qwen/Qwen2.5-3B-Instruct",
|
| 5 |
"adapter_path": "/kaggle/working/slow_consolidation_train/adapter",
|
| 6 |
"output_json": "/kaggle/working/slow_consolidation_train/kaggle_output.json",
|
| 7 |
+
"loss": 1.310697078704834,
|
| 8 |
"sft_source": "local",
|
| 9 |
"source_note": "M1 consolidation_sft.jsonl (lokal, §10.3)",
|
| 10 |
"dataset": "consolidation_sft.jsonl",
|
|
|
|
| 21 |
"num_train_epochs": 15.0,
|
| 22 |
"max_length": 1024,
|
| 23 |
"learning_rate": 0.0002,
|
| 24 |
+
"lora_r": 64,
|
| 25 |
+
"lora_alpha": 256,
|
| 26 |
"compute_dtype": "float16",
|
| 27 |
"amp": "disabled (fp16=False, bf16=False)",
|
| 28 |
"compute_capability": "sm_75",
|
| 29 |
"model_init_dtype": "float16",
|
| 30 |
"gpu": "Tesla T4",
|
| 31 |
+
"started_utc": "2026-10-19T17:21:52Z",
|
| 32 |
+
"finished_utc": "2026-10-19T17:24:37Z",
|
| 33 |
"library_versions": {
|
| 34 |
"transformers": "5.16.1",
|
| 35 |
"peft": "0.20.0",
|