OpenNMT/OpenNMT-py issues and pull requests

#2600 - download the model error

Issue - State: open - Opened by lemon1-ui 28 days ago

#2599 - Support for relative position embedding in the Encoder-Decoder attention (Context Attention)

Issue - State: open - Opened by frankang 2 months ago

#2598 - use gpu 1

Issue - State: closed - Opened by prigioni 4 months ago - 3 comments

#2597 - An error in model's partition and checkpoint's slice was detected

Issue - State: open - Opened by saltn9 5 months ago

#2596 - copy_attn causes RuntimeError: Expected all tensors to be on the same device, but found at least two devices, cuda:0 and cpu! (when checking argument for argument index in method wrapper_CUDA__index_select)

Issue - State: closed - Opened by aaaallleen 5 months ago - 3 comments

#2595 - End of support announcement

Pull Request - State: closed - Opened by vince62s 5 months ago

#2594 - all_reduce_and_rescale_tensors gives IndexError: list index out of range

Issue - State: closed - Opened by Garfounkel 5 months ago - 1 comment

#2593 - added the option to use BLEU and TER as a stopping criteria

Pull Request - State: closed - Opened by aaaallleen 5 months ago - 2 comments

#2592 - Update misc.py to prevent a runtime crash

Pull Request - State: closed - Opened by royshil 5 months ago - 1 comment

#2591 - High training/validation PPL observed with v3 compared to v2

Issue - State: closed - Opened by robertBrnnn 5 months ago - 3 comments

#2590 - Unable to use translate

Issue - State: closed - Opened by Aminoacid1226 6 months ago - 1 comment

#2589 - fixed masked flash attention

Pull Request - State: closed - Opened by l-k-11235 6 months ago - 1 comment

#2588 - free space for build, do not stop if a build is failing

Pull Request - State: closed - Opened by funboarder13920 6 months ago

#2587 - Bump jinja2 from 3.1.3 to 3.1.4 in /docs

Pull Request - State: closed - Opened by dependabot[bot] 7 months ago
Labels: dependencies

#2586 - [Bug - Translation server] - Missing `tgt`param in `translator.translate` method (allows some multilingual/seq2seq models to work properly)

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

#2585 - Fixed server not passing sequence `ref`/`tgt` to `translator.translate` method

Pull Request - State: open - Opened by medfreeman 7 months ago

#2584 - Supported languages

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

#2583 - Error when translating with scores (--with_scores) option enabled

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

#2582 - Issues with Custom SentencePiece Models and Pretrained Embeddings in Training

Issue - State: closed - Opened by HURIMOZ 7 months ago

#2581 - Canʻt get past Sentencepiece subword tokenization with pretrained embeddings

Issue - State: open - Opened by HURIMOZ 7 months ago

#2581 - Canʻt get past Sentencepiece subword tokenization with pretrained embeddings

Issue - State: closed - Opened by HURIMOZ 7 months ago

#2580 - Index out of range

Issue - State: closed - Opened by MSKantulu 7 months ago - 2 comments

#2580 - Index out of range

Issue - State: open - Opened by MSKantulu 7 months ago

#2579 - UnicodeDecodeError: 'utf-8' codec can't decode byte 0x80 in position 55: invalid start byte

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

#2579 - UnicodeDecodeError: 'utf-8' codec can't decode byte 0x80 in position 55: invalid start byte

Issue - State: open - Opened by fkurushin 7 months ago

#2578 - AttributeError: 'Namespace' object has no attribute 'block_ngram_repeat'

Issue - State: open - Opened by HURIMOZ 7 months ago

#2578 - AttributeError: 'Namespace' object has no attribute 'block_ngram_repeat'

Issue - State: closed - Opened by HURIMOZ 7 months ago

#2577 - Small changes in greed search

Pull Request - State: closed - Opened by l-k-11235 7 months ago

#2576 - add warm-up method for inference engines

Pull Request - State: closed - Opened by l-k-11235 8 months ago

#2576 - add warm-up method for inference engines

Pull Request - State: closed - Opened by l-k-11235 8 months ago

#2575 - Custom callbacks for metrics, saving checkpoints

Issue - State: open - Opened by Garfounkel 8 months ago - 4 comments

#2574 - OpenNMT v3.5.0 training fails using Multi headed attention

Issue - State: closed - Opened by RakshaPRao 8 months ago - 3 comments

#2573 - Reset rotary embeddings for chained inference

Pull Request - State: closed - Opened by l-k-11235 8 months ago

#2573 - Reset rotary embeddings for chained inference

Pull Request - State: open - Opened by l-k-11235 8 months ago

#2572 - new awq kernels paths

Pull Request - State: closed - Opened by vince62s 8 months ago

#2571 - (Again, but different) AssertionError: assert model_dim % head_count == 0

Issue - State: closed - Opened by James-Decatur 8 months ago - 2 comments

#2570 - bump 3.5.1

Pull Request - State: closed - Opened by vince62s 8 months ago

#2569 - Fixes

Pull Request - State: closed - Opened by vince62s 8 months ago

#2568 - Fix valid stats when we zero-out the prompt-loss

Pull Request - State: closed - Opened by l-k-11235 8 months ago

#2567 - Allow multiple response patterns in the insert_mask_before_placeholder transform

Pull Request - State: closed - Opened by l-k-11235 8 months ago

#2566 - v3.5 hotfix

Pull Request - State: closed - Opened by vince62s 9 months ago

#2565 - bump 3.5

Pull Request - State: closed - Opened by vince62s 9 months ago

#2564 - fix generation with large sequences when flash2 is False

Pull Request - State: closed - Opened by l-k-11235 9 months ago

#2563 - various fixes

Pull Request - State: closed - Opened by vince62s 9 months ago

#2562 - Device side assert triggered on AWQ Mistral converted model

Issue - State: closed - Opened by kdcyberdude 9 months ago - 2 comments

#2561 - Fix generation with large sequences

Pull Request - State: closed - Opened by l-k-11235 9 months ago - 1 comment

#2560 - Support for torch 2.2

Issue - State: closed - Opened by jakeBass 9 months ago - 4 comments

#2559 - NaN values when training big transformer model

Issue - State: closed - Opened by PC91 10 months ago - 1 comment

#2558 - fix spacing and fast rms not for training

Pull Request - State: closed - Opened by vince62s 10 months ago

#2557 - Fix bucket refilling in _score methode of Inference class

Pull Request - State: closed - Opened by l-k-11235 10 months ago

#2556 - How to use Huawei‘s NPU Ascend310 to install OpenNMT-py？

Issue - State: closed - Opened by michelleqyhqyh 10 months ago - 1 comment

#2555 - Translation API Not Working

Issue - State: closed - Opened by Keram-Yasin 10 months ago - 1 comment

#2554 - Speech to Text Toy Data Could Not Be Downloaded

Issue - State: closed - Opened by Keram-Yasin 10 months ago - 3 comments

#2553 - rolling ppl with sliding window

Pull Request - State: closed - Opened by l-k-11235 10 months ago

#2552 - misc + fix "\n" tokenization + phi-2 new layer names

Pull Request - State: closed - Opened by vince62s 10 months ago

#2551 - Bump jinja2 from 3.0.3 to 3.1.3 in /docs

Pull Request - State: closed - Opened by dependabot[bot] 10 months ago
Labels: dependencies

#2550 - Fixed OOM list out of range bug

Pull Request - State: open - Opened by alexis-allemann 10 months ago - 1 comment

#2550 - Fixed OOM list out of range bug

Pull Request - State: closed - Opened by alexis-allemann 10 months ago - 4 comments

#2549 - List index out of range in onmt.utils.distributed.all_reduce_and_rescale_tensors:51

Issue - State: open - Opened by alexis-allemann 10 months ago

#2549 - List index out of range in onmt.utils.distributed.all_reduce_and_rescale_tensors:51

Issue - State: open - Opened by alexis-allemann 10 months ago

#2548 - Support for Microsoft's Phi-2 model

Pull Request - State: closed - Opened by vince62s 10 months ago

#2547 - Add wikitext2 ppl benchmark with fixed context length in eval_llm

Pull Request - State: closed - Opened by l-k-11235 11 months ago

#2546 - Supported SentencePiece parameters

Issue - State: closed - Opened by PC91 11 months ago - 1 comment

#2545 - clean up opts

Pull Request - State: closed - Opened by vince62s 11 months ago

#2544 - fix #2329

Pull Request - State: closed - Opened by vince62s 11 months ago

#2543 - Enable scoring in inference engine py

Pull Request - State: closed - Opened by l-k-11235 11 months ago

#2542 - Restored masked scaled dot attention

Pull Request - State: closed - Opened by l-k-11235 11 months ago - 1 comment

#2541 - Error evaluating LM-prior checkpoint:

Issue - State: closed - Opened by anthdr 11 months ago - 1 comment

#2540 - remove llm-awq dependancy (conflict with autoawq)

Pull Request - State: closed - Opened by vince62s 11 months ago

#2539 - use flash_attn_with_kvcache for faster inference

Pull Request - State: closed - Opened by vince62s 11 months ago

#2538 - Fixmask

Pull Request - State: closed - Opened by vince62s 11 months ago

#2537 - Defaultflash

Pull Request - State: closed - Opened by vince62s 11 months ago

#2536 - Extend llamalike-converter

Pull Request - State: closed - Opened by vince62s 11 months ago

#2535 - MoE for mixtral 8x7b

Pull Request - State: open - Opened by vince62s 11 months ago

#2535 - MoE for mixtral 8x7b

Pull Request - State: closed - Opened by vince62s 11 months ago

#2534 - fix multi gpu valid for data_parallel

Pull Request - State: closed - Opened by vince62s 12 months ago

#2534 - fix multi gpu valid for data_parallel

Pull Request - State: closed - Opened by vince62s 12 months ago

#2533 - provide a test for rotary embeddings with left padding

Pull Request - State: closed - Opened by l-k-11235 12 months ago

#2533 - provide a test for rotary embeddings with left padding

Pull Request - State: open - Opened by l-k-11235 12 months ago

#2532 - Apply the attention mask in all decoding steps (LM inference)

Pull Request - State: closed - Opened by l-k-11235 12 months ago

#2531 - Distributed inference of 70B awq model

Pull Request - State: closed - Opened by vince62s 12 months ago

#2531 - Distributed inference of 70B awq model

Pull Request - State: closed - Opened by vince62s 12 months ago

#2530 - Add some official Docker images

Pull Request - State: closed - Opened by francoishernandez 12 months ago - 4 comments

#2529 - fix bnb loading

Pull Request - State: closed - Opened by vince62s 12 months ago

#2528 - Bug when training encoder-decoder models

Issue - State: closed - Opened by JOHW85 12 months ago - 1 comment

#2527 - Fixconv

Pull Request - State: closed - Opened by vince62s 12 months ago

#2526 - fix docify

Pull Request - State: closed - Opened by vince62s 12 months ago

#2525 - left padding for LM inference

Pull Request - State: closed - Opened by l-k-11235 12 months ago - 1 comment

#2524 - update to skipinit weights with bnb

Pull Request - State: closed - Opened by vince62s 12 months ago

#2523 - Set dynamic max length per batch

Pull Request - State: closed - Opened by vince62s 12 months ago

#2522 - Error message of `SequenceTooLongError`

Issue - State: closed - Opened by PC91 12 months ago - 1 comment

#2521 - Check if input tensor is empty before reducing and rescaling

Pull Request - State: closed - Opened by PC91 almost 1 year ago - 3 comments

GitHub / OpenNMT/OpenNMT-py issues and pull requests