google gemma_pytorch issues

google / gemma_pytorch

The official PyTorch implementation of Google's Gemma models

https://ai.google.dev/gemma

Apache License 2.0

5.19k stars 492 forks source link

issues

Newest

Newest Most commented Recently updated Oldest Least commented Least recently updated

Output with higher max_length is repetition of base text

#19 azrael05 opened 5 months ago
7
Update config.py

#18 Khajaamee455 closed 4 months ago
2
Updated ClassMD

#17 Masomabayat closed 5 months ago
2
Update xla_model_parallel.py

#15 eltociear closed 5 months ago
3
RuntimeError: probability tensor contains either `inf`, `nan` or element < 0

#13 2579356425 closed 5 months ago
6
Are there reserved/unused tokens for developers?

#12 Qubitium closed 4 months ago
3
MPS (Apple Silicon) Support

#11 dsanmart opened 5 months ago
2
why some prompt doesn't work, the hidden_states will be nan after GemmaModel.forward

#10 vupjing opened 5 months ago
8
Loading torch checkpoint with weights_only set to True

#9 michaelmoynihan closed 5 months ago
0
how to finetune with gemma model？

#8 runningabcd opened 5 months ago
8
Quantised weights are bfloat16 not int8

#7 dsanmart closed 5 months ago
3
Add utility to convert string to boolean type to fix quant parse arg

#6 nakkapeddi closed 5 months ago
3
--quant always returns True

#5 nakkapeddi closed 5 months ago
5
A web runtime supported version of gemma is really needed and high value

#4 Zwe1 opened 5 months ago
2
RuntimeError: at::cuda::blas::gemm: not implemented for struct c10::BFloat16

#3 dhchenx closed 5 months ago
9
Inconsistencies in Reported Dimensions and Configuration Files

#2 fvarno closed 5 months ago
2
`torch.load` without `weights_only` parameter is unsafe

#1 kit1980 closed 5 months ago
2