Ecosyste.ms: Issues

An open API service for providing issue and pull request metadata for open source projects.

GitHub / google/gemma_pytorch issues and pull requests

#77 - Failed to cancel access request

Issue - State: open - Opened by lucassunalt 20 days ago - 1 comment
Labels: stat:awaiting response

#76 - Add required world_size and rank to GemmaDecodeLayer init

Pull Request - State: closed - Opened by DavidRV00 about 1 month ago - 2 comments

#73 - Non causal sliding window mask ?

Issue - State: closed - Opened by Optimox 3 months ago - 3 comments

#71 - Inconsistent 'query_pre_attn_scalar' Setting Between 9B and 27B Models

Issue - State: open - Opened by kiddj 6 months ago - 2 comments
Labels: bug, stat:awaiting response

#70 - Hope to See the Source Code of Gemma2 Version

Issue - State: closed - Opened by thefreeman007 7 months ago - 1 comment
Labels: type:support

#69 - Remove unused imports

Pull Request - State: closed - Opened by neurosnap 7 months ago - 1 comment

#68 - Fix downcasting and upcasting similar to https://github.com/google/ge…

Pull Request - State: closed - Opened by michaelmoynihan 7 months ago - 1 comment

#67 - Fix downcasting and upcasting

Pull Request - State: closed - Opened by danielhanchen 7 months ago - 1 comment

#66 - Supporting Gemma V2

Pull Request - State: closed - Opened by michaelmoynihan 7 months ago - 1 comment

#65 - Update run_xla.py

Pull Request - State: closed - Opened by michaelmoynihan 7 months ago

#64 - gemma-2b-it-pytorch on tpu v5p

Issue - State: closed - Opened by shungcp 7 months ago - 1 comment

#63 - Modify SentencePiece function calls.

Pull Request - State: closed - Opened by texasmichelle 8 months ago - 1 comment

#62 - Change return to raise in `get_model_config`.

Pull Request - State: closed - Opened by texasmichelle 8 months ago - 1 comment

#61 - when to support RecurrentGemma?

Issue - State: closed - Opened by Mddct 9 months ago - 1 comment
Labels: enhancement

#60 - Gemma finetuning formatting

Issue - State: closed - Opened by mostafamdy 9 months ago - 3 comments
Labels: type:support

#59 - fix missing torch in requirment

Pull Request - State: closed - Opened by Mddct 9 months ago - 1 comment

#58 - Add CodeGemma and HF pointers

Pull Request - State: closed - Opened by osanseviero 10 months ago - 1 comment

#57 - early stop when all sequence reach EOS

Pull Request - State: open - Opened by je1lee 10 months ago - 3 comments

#56 - Memory saving loading weight for non-quant models

Pull Request - State: closed - Opened by KaneGreen 10 months ago - 5 comments

#55 - Prepare model for deployment to Private Vertex AI endpoint

Issue - State: closed - Opened by BriianPowell 10 months ago - 5 comments
Labels: type:support

#54 - Update xla_model_parallel.py

Pull Request - State: closed - Opened by ya0guang 10 months ago - 2 comments

#53 - Error when run docker/Dockerfile

Issue - State: closed - Opened by Cguanqin 10 months ago - 3 comments
Labels: type:support, stat:awaiting response

#52 - How to use gemma for multi-round conversations

Issue - State: closed - Opened by ranck626 10 months ago - 4 comments
Labels: type:support, stat:awaiting response

#51 - How to save memory when loading weights?

Issue - State: closed - Opened by KaneGreen 10 months ago - 9 comments
Labels: bug

#50 - Unable to reproduce MATH resulst

Issue - State: open - Opened by wenhuchen 10 months ago - 2 comments
Labels: type:support, stat:awaiting internal

#49 - fix: raise Exception

Pull Request - State: closed - Opened by leowzz 10 months ago - 2 comments

#48 - Is it possible to load 7b-it using quantization config

Issue - State: closed - Opened by aliasneo1 10 months ago - 1 comment
Labels: enhancement

#47 - Error when running Gemma inference on GPU

Issue - State: closed - Opened by LarryHawkingYoung 10 months ago - 3 comments
Labels: type:support, stat:awaiting response

#46 - rm fairescale

Pull Request - State: closed - Opened by Mon-ius 10 months ago - 7 comments

#45 - I got empty result while using 7b-it model

Issue - State: closed - Opened by egbertwong 11 months ago - 4 comments
Labels: type:support

#44 - Document the existence of 99 unused tokens in the tokenizer

Pull Request - State: closed - Opened by Qubitium 11 months ago - 1 comment

#43 - fix(temperature): allow passing 0 or None as the temperature parameter

Pull Request - State: closed - Opened by joselpart 11 months ago - 3 comments

#42 - Can't disable sampling

Issue - State: closed - Opened by joselpart 11 months ago
Labels: bug

#41 - Is max_position_embeddings=8096 neccessary in 2b model?

Issue - State: closed - Opened by agiwave 11 months ago - 5 comments
Labels: type:support

#40 - Auto-labels 'Gemma' on 'gemma' issues/PRs.

Pull Request - State: closed - Opened by shmishra99 11 months ago - 1 comment

#39 - Objectivity

Issue - State: closed - Opened by o6uoq 11 months ago
Labels: type:support

#38 - How to fine-tune Gemma with pytorch?

Issue - State: closed - Opened by solitude-alive 11 months ago - 2 comments
Labels: duplicate

#37 - Gemma fixes - gelu

Pull Request - State: closed - Opened by danielhanchen 11 months ago - 4 comments

#36 - Torch implementation now same as JAX

Pull Request - State: closed - Opened by thebraingen 11 months ago - 1 comment

#35 - Implementation now equals JAX

Pull Request - State: closed - Opened by thebraingen 11 months ago - 1 comment

#34 - Add instructions to download from Hugging Face Hub

Pull Request - State: closed - Opened by osanseviero 11 months ago - 1 comment

#33 - Inconsistency between PyTorch and JAX implementation

Issue - State: closed - Opened by aboros98 11 months ago - 2 comments
Labels: enhancement

#32 - "--output_len" argument ignored

Pull Request - State: closed - Opened by k-nar 11 months ago - 1 comment

#31 - not found weight file

Issue - State: closed - Opened by Cguanqin 11 months ago - 5 comments
Labels: type:support, stat:awaiting response

#30 - is it possible to convert gemma_pytorch to onnx to tflite?

Issue - State: closed - Opened by nyadla-sys 11 months ago - 4 comments
Labels: type:support, stat:awaiting response

#29 - [Question] Embeddings normalization by sqrt(hidden_size)

Issue - State: closed - Opened by Andrei-Aksionov 11 months ago - 4 comments
Labels: type:support

#26 - After deplyed google/gemma-7b-it, there always is error response.

Issue - State: closed - Opened by ydh10002023 11 months ago - 10 comments
Labels: bug

#25 - Cannot run on v4-16 worker 0 TPU VM: "Failed to get global TPU topology"

Issue - State: closed - Opened by markusheimerl 11 months ago - 6 comments
Labels: type:support

#24 - always loss nan while finetune a few step, wether fp32 or fp16

Issue - State: closed - Opened by yongzhuo 11 months ago - 1 comment
Labels: type:support

#23 - keras finetuning and inference examples uploaded

Pull Request - State: closed - Opened by r-gheda 11 months ago - 2 comments

#22 - H

Issue - State: closed - Opened by ZainBinTariq7 11 months ago - 1 comment
Labels: type:support

#21 - Changed <2B or 7B> to <2b or 7b> in README

Pull Request - State: closed - Opened by r-gheda 11 months ago

#20 - Changes <2B or 7B> option to <2b or 7b> in README

Pull Request - State: closed - Opened by r-gheda 11 months ago - 1 comment

#19 - Output with higher max_length is repetition of base text

Issue - State: closed - Opened by azrael05 11 months ago - 9 comments
Labels: type:support, stat:awaiting response

#18 - Update config.py

Pull Request - State: closed - Opened by Khajaamee455 11 months ago - 2 comments

#17 - Updated ClassMD

Pull Request - State: closed - Opened by Masomabayat 11 months ago - 2 comments

#15 - Update xla_model_parallel.py

Pull Request - State: closed - Opened by eltociear 11 months ago - 3 comments

#13 - RuntimeError: probability tensor contains either `inf`, `nan` or element < 0

Issue - State: closed - Opened by 2579356425 11 months ago - 6 comments
Labels: duplicate

#12 - Are there reserved/unused tokens for developers?

Issue - State: closed - Opened by Qubitium 11 months ago - 3 comments
Labels: type:support

#11 - MPS (Apple Silicon) Support

Issue - State: open - Opened by dsanmart 11 months ago - 3 comments
Labels: enhancement

#10 - why some prompt doesn't work, the hidden_states will be nan after GemmaModel.forward

Issue - State: closed - Opened by vupjing 11 months ago - 11 comments
Labels: bug, stat:awaiting response

#9 - Loading torch checkpoint with weights_only set to True

Pull Request - State: closed - Opened by michaelmoynihan 11 months ago

#8 - how to finetune with gemma model？

Issue - State: closed - Opened by runningabcd 11 months ago - 10 comments
Labels: type:support, stat:awaiting response

#7 - Quantised weights are bfloat16 not int8

Issue - State: closed - Opened by dsanmart 11 months ago - 3 comments
Labels: type:support

#6 - Add utility to convert string to boolean type to fix quant parse arg

Pull Request - State: closed - Opened by nakkapeddi 11 months ago - 3 comments

#5 - --quant always returns True

Issue - State: closed - Opened by nakkapeddi 11 months ago - 5 comments
Labels: bug

#4 - A web runtime supported version of gemma is really needed and high value

Issue - State: closed - Opened by Zwe1 11 months ago - 4 comments
Labels: enhancement

#3 - RuntimeError: at::cuda::blas::gemm: not implemented for struct c10::BFloat16

Issue - State: closed - Opened by dhchenx 11 months ago - 9 comments
Labels: bug

#2 - Inconsistencies in Reported Dimensions and Configuration Files

Issue - State: closed - Opened by fvarno 11 months ago - 2 comments
Labels: type:support

#1 - `torch.load` without `weights_only` parameter is unsafe

Issue - State: closed - Opened by kit1980 11 months ago - 2 comments
Labels: bug