dereklll comments

Repositories
Issues
Comments

Results 6 comments of


                                            dereklll

Can I delete line 91 "self.check_op = tf.add_check_numerics_ops()" ??

any solution?

I may have found the reason why the model can't be trained.

I‘ll give you guy a sixsixsix.

I may have found the reason why the model can't be trained.

oh,no. I want my sixsixsix back.

I may have found the reason why the model can't be trained.

> @dereklll > 意思是一开始有效，后面又失效了？你这样正样本loss全部为0，被mask了，total loss只有负样本的loss

ggml_new_tensor_impl: not enough space in the context's memory pool

Same here... any solutions already???

类似 llama.cpp的加速？

> (alpaca_env) chunzhamini@chunzhamini llama.cpp % ./main -m ./zh-models/baichuan/Baichuan2-13B-Chat-ggml-model-q4_0.bin -p '从前有一只小狐狸，他' --temp 0 -ngl 1 Log start main: warning: changing RoPE frequency base to 0 (default 10000.0) main: warning: scaling RoPE...