Skip to content

🧠 两小时从 0 训出你的 64M 大模型。Train your own 64M LLM from scratch in 2 hours.

update 0-eval_pretrain.py

gongjyb00a3b5ff750
diff --git a/0-eval_pretrain.py b/0-eval_pretrain.py
index 0fa9710..32f1dda 100644
--- a/0-eval_pretrain.py
+++ b/0-eval_pretrain.py
@@ -16,7 +16,7 @@ def count_parameters(model):
def init_model(lm_config):
- tokenizer = AutoTokenizer.from_pretrained('./model',
+ tokenizer = AutoTokenizer.from_pretrained('./model/minimind_tokenizer',
trust_remote_code=True, use_fast=False)
model_from = 1 # 1从权重,2用transformers