Upload config_train.json with huggingface_hub

config_train.json

@@ -0,0 +1,26 @@
+{
+  "input_csv": "data/bdx_codes/train.csv",
+  "text_col": "text",
+  "label_col": "dp_code_tronque",
+  "output_dir": "checkpoints/camembert_dp_ft_top40_bdx",
+  "pretrained": "almanach/camembert-bio-base",
+  "epochs": 20,
+  "batch_size": 8,
+  "lr": 2e-05,
+  "fp16": true,
+  "bf16": false,
+  "warmup_ratio": 0.1,
+  "weight_decay": 0.01,
+  "max_length": 512,
+  "val_frac": 0.1,
+  "seed": 42,
+  "eval_every": 500,
+  "save_total_limit": 2,
+  "gradient_accumulation_steps": 4,
+  "do_chunk": true,
+  "chunk_size": 480,
+  "chunk_stride": 64,
+  "mlflow_run_name": "finetuning_top40_bdx",
+  "n_train": 8407,
+  "n_val": 956
+}