matrix comments

Results 72 comments of


                                            matrix

Multilingual or Chinese version plan?

hope to provide a large Chinese model. Thank you very much.

Multilingual or Chinese version plan?

With the rapid development of natural language processing, the slower the release of the relevant version, its value will drop rapidly, and hope that your work can help more people's...

For 30B LLama model, can server be supported by configuring mesh_dims on tpu v3-8 (128g)? I tried 8,1 and 4,1 but they don't seem to work.

Currently, there is still no work.

When I use the following code on tpuvm and use model.generate() to infer, the speed is very slow. It seems that the tpu is not used. What is the problem?

Thank you very much for your reply, I tried it, it is indeed effective In addition, It reports OOM on the V3-8TPU to use MT5-XXL. do you have any suggestions?...

When I use the following code on tpuvm and use model.generate() to infer, the speed is very slow. It seems that the tpu is not used. What is the problem?

great! Thank you very much for your suggestion. I will try it next

When I use the following code on tpuvm and use model.generate() to infer, the speed is very slow. It seems that the tpu is not used. What is the problem?

ok, Does this method also support the XXL model on the TPU V3-8?

When I use the following code on tpuvm and use model.generate() to infer, the speed is very slow. It seems that the tpu is not used. What is the problem?

ok, I am very grateful for your suggestion, I plan to try and experiment further

When I use the following code on tpuvm and use model.generate() to infer, the speed is very slow. It seems that the tpu is not used. What is the problem?

When I load it with this model"ClueAI/ChatYuan-large-v1", the following error will occur. How to solve this problem? ```shell Some weights of the model checkpoint at ClueAI/ChatYuan-large-v1 were not used when...

When I use the following code on tpuvm and use model.generate() to infer, the speed is very slow. It seems that the tpu is not used. What is the problem?

> #16148 (comment) while i try, error occur, How to solve this problem? ```python model, params = FlaxT5ForConditionalGeneration.from_pretrained("ClueAI/ChatYuan-large-v1", _do_init=False) ``` ```python OSError: ClueAI/ChatYuan-large-v1 does not appear to have a file...

When I use the following code on tpuvm and use model.generate() to infer, the speed is very slow. It seems that the tpu is not used. What is the problem?

i try it, and How to configure "max_length", "top_k", "do_sample" and other parameters with this ? https://github.com/sanchit-gandhi/codesnippets/blob/main/pmap_flaxmt5_generate.ipynb