From 47ba18a4f9d50807744cb3f4c64f3992cb110d66 Mon Sep 17 00:00:00 2001 From: Colin <> Date: Wed, 13 Aug 2025 22:44:09 +0800 Subject: [PATCH] Refine base model. --- wit/train.py | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-) diff --git a/wit/train.py b/wit/train.py index 9538dc1..5fab291 100644 --- a/wit/train.py +++ b/wit/train.py @@ -23,7 +23,7 @@ if __name__ == "__main__": conf.train_batch_size = 32 conf.val_batch_size = 2 conf.num_proc = 8 - conf.max_epochs = 1000 + conf.max_epochs = 30 conf.strategy = "auto" conf.resume_from_ckpt_path = None conf.seed = 42 @@ -40,8 +40,8 @@ if __name__ == "__main__": conf.dataset.meaning.val_mask_idx = [0, 0, -1] config.vocab_size = 32 - config.hidden_size = 256 # 128 1024 2048 32 - config.intermediate_size = 512 + config.hidden_size = 128 # 128 1024 2048 32 + config.intermediate_size = 256 config.num_hidden_layers = 3 # 6 12 24 3 config.num_attention_heads = 4 # 8 8 16