Ecosyste.ms: Issues

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

GitHub / Ki6an/fastT5 issues and pull requests

#77 - Does UMT5 Supports?

Issue - State: open - Opened by jhpassion0621 5 months ago - 2 comments

#74 - Dead kernel

Issue - State: open - Opened by benzom 11 months ago

#71 - compat transformers 4.28 & onnxruntime 1.14

Pull Request - State: open - Opened by mzurek-teldoc about 1 year ago

#70 - Segmentation fault (core dumped)

Issue - State: open - Opened by lisi-github about 1 year ago - 5 comments

#69 - GPU support

Issue - State: open - Opened by p-christ over 1 year ago

#68 - flan-t5 support

Issue - State: open - Opened by loretoparisi over 1 year ago - 5 comments

#67 - Issue with onnxruntime

Issue - State: open - Opened by GenVr over 1 year ago - 4 comments

#66 - In the source code, the use of attention_mask is contradictory

Issue - State: open - Opened by ghozn over 1 year ago - 1 comment

#65 - No such file or directory: '/content/encoder.embed_tokens.weight'

Issue - State: open - Opened by alexfdo over 1 year ago - 3 comments

#64 - fastt5 not working with FastAPI gunicorn and docker

Issue - State: open - Opened by kklivil over 1 year ago - 1 comment

#63 - Upgrade ONNX runtime

Issue - State: open - Opened by dandiep over 1 year ago - 1 comment

#61 - M2M100 to ONNX

Issue - State: open - Opened by sorgfresser almost 2 years ago

#60 - Thank You and Demo Running in the Browser

Issue - State: closed - Opened by praeclarum almost 2 years ago - 2 comments

#59 - Decoder's encoder_attention_mask input should be called decoder_attention_mask

Issue - State: closed - Opened by praeclarum almost 2 years ago - 1 comment

#58 - Is it suit for other translation model like "Helsinki-NLP / opus-mt-en-de"?

Issue - State: closed - Opened by xyx361100238 almost 2 years ago - 1 comment

#57 - Mt5 model loading fails

Issue - State: open - Opened by OriAlpha about 2 years ago - 11 comments

#56 - Fails to convert T0-3B

Issue - State: open - Opened by redthing1 about 2 years ago - 2 comments

#55 - Support for py3.10

Issue - State: closed - Opened by adityay121 about 2 years ago - 1 comment

#54 - Not able to run onnx model of allen ai t5 small on GPU

Issue - State: closed - Opened by RKGhule about 2 years ago - 1 comment

#53 - Failed to create CUDAExecutionProvider

Issue - State: closed - Opened by ekaterinatretyak about 2 years ago - 1 comment

#52 - Unable to retrieve hidden_states

Issue - State: open - Opened by vsoesanto about 2 years ago - 2 comments

#51 - Electra model loading into fastT5

Issue - State: closed - Opened by OriAlpha about 2 years ago

#50 - Conversion of decoder with past_key_values to float16.

Issue - State: closed - Opened by VikasOjha666 about 2 years ago - 5 comments

#49 - Observing difference in outputs from decoder with IO bindings.

Issue - State: closed - Opened by VikasOjha666 over 2 years ago - 10 comments

#48 - quantize_dynamic() got an unexpected keyword argument 'activation_type'

Issue - State: closed - Opened by katkardhanraj over 2 years ago - 1 comment

#46 - Support latest transformers

Pull Request - State: closed - Opened by Ki6an over 2 years ago

#45 - offline install error

Issue - State: open - Opened by JingxinLee over 2 years ago - 1 comment

#44 - how to fix when convert model mT5 with max_length = 512

Issue - State: open - Opened by batman-do over 2 years ago - 3 comments

#43 - How to fix when run convert model mT5

Issue - State: closed - Opened by batman-do over 2 years ago - 1 comment

#42 - GPU support for fastT5

Issue - State: closed - Opened by rupeshpoojary97 over 2 years ago - 3 comments

#41 - Implemented the code for BART

Issue - State: closed - Opened by siddharth-sharma7 over 2 years ago

#40 - Incompatible with transformers>=4.16

Issue - State: closed - Opened by nbroad1881 over 2 years ago - 1 comment

#39 - Accuracy hit when using fastT5

Issue - State: open - Opened by mbuotidem over 2 years ago

#38 - the method get_onnx_model() should not need the path to original model

Issue - State: open - Opened by piegu over 2 years ago - 2 comments

#37 - Cuda support

Pull Request - State: closed - Opened by fmaoro over 2 years ago - 3 comments

#36 - Is fastT5 deepspeed compatible?

Issue - State: closed - Opened by jhillhouse92 over 2 years ago

#35 - Explicit mt5 support

Pull Request - State: closed - Opened by sam-writer over 2 years ago

#34 - GPU Optimization

Issue - State: open - Opened by ashissamal over 2 years ago - 7 comments
Labels: enhancement

#33 - Errors when loading saved onnx files

Issue - State: closed - Opened by itaim over 2 years ago - 10 comments

#32 - small onnx optimizations

Pull Request - State: closed - Opened by sam-writer over 2 years ago - 4 comments

#31 - Where is model saved after it is converted to ONNX?

Issue - State: closed - Opened by Oxi84 over 2 years ago - 1 comment

#30 - add `psutil` to deps in `setup.py`

Pull Request - State: closed - Opened by aseifert over 2 years ago - 1 comment

#29 - add custom models folder support in get_onnx_model

Pull Request - State: closed - Opened by WarrierRajeev over 2 years ago - 3 comments

#28 - get_onnx_model fails

Issue - State: closed - Opened by WarrierRajeev over 2 years ago - 1 comment

#27 - Updating fastT5?

Issue - State: closed - Opened by piegu over 2 years ago - 6 comments

#26 - default to 0 threads so ORT can choose

Pull Request - State: closed - Opened by sam-writer over 2 years ago

#25 - ONNX Runtime Session Defaults Discrepancy

Issue - State: closed - Opened by sam-writer over 2 years ago - 1 comment

#24 - Error from export_and_get_onnx_model()

Issue - State: closed - Opened by PhaneendraGunda over 2 years ago - 3 comments

#23 - OnnxT5 slower than Pytorch

Issue - State: open - Opened by GRIGORR over 2 years ago - 16 comments

#22 - How to use numpy inputs?

Issue - State: closed - Opened by Anku5hk almost 3 years ago - 1 comment

#21 - Allow use of private model hub models

Pull Request - State: closed - Opened by sam-writer almost 3 years ago - 2 comments

#20 - How can we use in mT5

Issue - State: closed - Opened by sainathpawar almost 3 years ago - 2 comments

#19 - A few improvements

Pull Request - State: closed - Opened by kagrze about 3 years ago - 1 comment

#18 - forward() got an unexpected keyword argument 'cross_attn_head_mask'

Issue - State: closed - Opened by mrm8488 about 3 years ago - 20 comments

#17 - Error: Duplicate definition of name (pkv_11) using fine-tuned T5

Issue - State: closed - Opened by mrm8488 about 3 years ago - 1 comment

#14 - CPU-inference T5Encoder-XL slower than PyTorch

Issue - State: closed - Opened by CodyLDA about 3 years ago - 3 comments

#13 - Question about implementation details of `past_key_values`

Issue - State: closed - Opened by ryangawei about 3 years ago

#12 - Issue with quantize()

Issue - State: closed - Opened by vikramelango about 3 years ago - 1 comment

#11 - t5-11b out of memory/FileNotFoundError

Issue - State: closed - Opened by ViktorThink about 3 years ago - 5 comments

#10 - failed with output shape error

Issue - State: closed - Opened by rossbucky about 3 years ago - 1 comment

#9 - Problems with T5 & onnxruntime

Issue - State: closed - Opened by GenVr about 3 years ago - 4 comments

#8 - can quantized onnx t5 run on GPU

Issue - State: closed - Opened by rossbucky about 3 years ago - 3 comments

#7 - Different behaviour when extending this project to Bart

Issue - State: open - Opened by HenryDashwood about 3 years ago - 35 comments

#6 - OnnxT5 creates potentially unneccessary objects

Issue - State: closed - Opened by tobigue over 3 years ago

#5 - Not all outputs have names

Issue - State: closed - Opened by tobigue over 3 years ago - 2 comments

#4 - Not working

Issue - State: closed - Opened by Anku5hk over 3 years ago - 2 comments

#3 - Error with "kiri-ai/t5-base-qa-summary-emotion" model

Issue - State: closed - Opened by orendar over 3 years ago - 2 comments

#2 - mt5 and Neural Machine Trnslation

Issue - State: closed - Opened by seekingdeep over 3 years ago

#1 - Getting runtime error.

Issue - State: closed - Opened by abhinavsp0730 over 3 years ago - 4 comments
Labels: good first issue