Ecosyste.ms: Issues

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

GitHub / inseq-team/inseq issues and pull requests

#289 - Fix LLaMA 3.2, add `clean_special_chars`

Pull Request - State: closed - Opened by gsarti 25 days ago

#288 - How to set `max_length` in generation?

Issue - State: closed - Opened by Ningshiqi 26 days ago - 1 comment
Labels: question

#287 - Model LLaMA 3.2 not suported

Issue - State: closed - Opened by Ningshiqi 26 days ago - 4 comments
Labels: bug

#286 - CUDA OOM with mistralai/Mistral-7B-Instruct-v0.3

Issue - State: closed - Opened by Ningshiqi 26 days ago - 3 comments
Labels: bug

#285 - Does Inseq support attributions with example granularity as in Captum's few-shot?

Issue - State: open - Opened by Ningshiqi 27 days ago - 7 comments
Labels: question

#284 - `treescope` visualization for `attribute_context`

Pull Request - State: closed - Opened by gsarti 3 months ago

#283 - `treescope` support and new visualizations

Pull Request - State: closed - Opened by gsarti 3 months ago - 2 comments

#282 - Improve attribution viz and `SliceAggregator`

Pull Request - State: closed - Opened by gsarti 4 months ago

#281 - Show only part of the input when calling show() method

Issue - State: closed - Opened by RibinMTC 4 months ago - 2 comments
Labels: question

#280 - Fix multi-gpu inference and duplicate BOS issue for decoder-only

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

#279 - CUDA out of memory issue on long context

Issue - State: open - Opened by acDante 5 months ago - 4 comments
Labels: question

#278 - Forced generations with decoder-only models must start with input texts

Issue - State: closed - Opened by acDante 5 months ago - 1 comment
Labels: question

#277 - Generating explanations

Issue - State: closed - Opened by SVC04 5 months ago
Labels: question

#276 - Expected all tensors to be on the same device multi-gpu

Issue - State: closed - Opened by saxenarohit 6 months ago - 7 comments
Labels: bug

#275 - Add possibility to skip special tokens during attribution

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

#274 - `attribute_context` Error when output contains emoji

Issue - State: open - Opened by Betswish 6 months ago
Labels: bug

#273 - Save tensors in lower precision

Pull Request - State: closed - Opened by LuukSuurmeijer 6 months ago - 3 comments

#272 - Can't run inseq with batch_size>1 for methods that use gradients (e.g. integrated gradients)

Issue - State: closed - Opened by issam9 6 months ago - 2 comments
Labels: bug

#271 - Question about start token?

Issue - State: closed - Opened by FYYFU 6 months ago - 7 comments
Labels: question

#270 - Aggregate the attribution for each input sentence with LLama2 model

Issue - State: closed - Opened by acDante 6 months ago - 4 comments
Labels: question

#269 - Fix nans in attention, add pad in tokenizer if missing

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

#268 - Why sometimes the target attribution can be 'nan'?

Issue - State: closed - Opened by frankdarkluo 7 months ago - 7 comments
Labels: question

#267 - Support Value Zeroing for non-eager attention types

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

#266 - Issue with the implementation of Value Zeroing

Issue - State: closed - Opened by hmohebbi 7 months ago - 2 comments
Labels: bug

#265 - Does inseq work with Mistral-7B or Llama-2-7B models?

Issue - State: closed - Opened by hadiasghari 7 months ago - 3 comments
Labels: question

#264 - Add support for `device_map` use

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

#263 - Can I load the inseq model onto multiple GPUs?

Issue - State: closed - Opened by frankdarkluo 7 months ago - 5 comments
Labels: question

#262 - Update lime.py

Pull Request - State: open - Opened by yuzhaouoe 7 months ago - 1 comment

#261 - Which aggregation function is better?

Issue - State: closed - Opened by MarshtompCS 7 months ago - 1 comment
Labels: question

#260 - Long input document is truncated when calculating attribution scores

Issue - State: closed - Opened by acDante 8 months ago - 3 comments
Labels: question

#259 - Fix URL to arXiv

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

#258 - Various fixes to `attribute-context`

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

#257 - Update attribute_context.py

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

#256 - Sanitize strings from markup upon Rich printing

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

#255 - rich.errors.MarkupError for chat template

Issue - State: closed - Opened by mohsenfayyaz 9 months ago - 2 comments
Labels: bug

#254 - Rollout aggregation function

Pull Request - State: open - Opened by gsarti 9 months ago

#253 - Fix casting step outputs to device

Pull Request - State: closed - Opened by gsarti 9 months ago - 3 comments

#251 - Test failed on CUDA enabled platform

Issue - State: closed - Opened by xuan25 9 months ago
Labels: bug

#250 - Feature/reagent branch for ReAGent

Pull Request - State: closed - Opened by casszhao 9 months ago - 7 comments
Labels: enhancement

#249 - Dev QOL updates

Pull Request - State: closed - Opened by gsarti 9 months ago
Labels: dev qol

#248 - [Summary] Add step functions

Issue - State: open - Opened by gsarti 9 months ago
Labels: enhancement, help wanted, good first issue, summary

#247 - Fix `aggregate_contiguous`

Pull Request - State: closed - Opened by gsarti 10 months ago
Labels: bug

#246 - Producing heatmap visualizations with SubwordAggregator and SequenceAttributionAggregator

Issue - State: closed - Opened by nfelnlp 10 months ago - 1 comment
Labels: bug

#245 - `attribute` speed and memory optimizations

Pull Request - State: closed - Opened by gsarti 10 months ago - 1 comment
Labels: enhancement

#242 - Move documentation to `mkdocs`

Issue - State: open - Opened by gsarti 11 months ago
Labels: documentation, enhancement, long shot

#241 - Add a CHANGELOG.md file

Issue - State: closed - Opened by gsarti 11 months ago
Labels: enhancement

#240 - CUDA out of memory error.

Issue - State: closed - Opened by SVC04 11 months ago - 2 comments
Labels: question

#239 - Contrastive explanations

Issue - State: closed - Opened by JanAthmer 11 months ago - 4 comments
Labels: question

#238 - Multi-device support for attribution

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

#237 - `attribute-context` CLI command

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

#236 - Bump dependencies, update version and readme

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

#235 - Add optional jax group to enforce compatible jaxlib version.

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

#234 - Module 'jax' has no attribute 'typing'

Issue - State: closed - Opened by carschno 12 months ago - 3 comments
Labels: bug

#233 - Minor fixes

Pull Request - State: closed - Opened by gsarti about 1 year ago

#232 - Utilizing Inseq

Issue - State: closed - Opened by SVC04 about 1 year ago - 18 comments
Labels: question

#231 - Update tutorial to contrastive attribution changes

Pull Request - State: closed - Opened by gsarti about 1 year ago
Labels: bug

#230 - Expected all tensors to be on the same device

Issue - State: closed - Opened by saxenarohit about 1 year ago - 3 comments
Labels: bug, to be investigated

#229 - Fix command for installing pre-commit hooks.

Pull Request - State: closed - Opened by carschno about 1 year ago

#228 - Attributed behavior for contrastive step functions

Pull Request - State: closed - Opened by gsarti about 1 year ago
Labels: enhancement

#227 - Remove `max_input_length` from `model.encode`

Pull Request - State: closed - Opened by gsarti about 1 year ago

#226 - Migrate from torchtyping to jaxtyping

Pull Request - State: closed - Opened by carschno about 1 year ago - 1 comment

#225 - Migrate to `ruff format`

Pull Request - State: closed - Opened by gsarti about 1 year ago

#224 - Remove contrast_target_prefixes from contrastive step functions

Pull Request - State: closed - Opened by gsarti about 1 year ago

#223 - Step functions fixes, add `in_context_pvi`

Pull Request - State: closed - Opened by gsarti about 1 year ago

#222 - Add Sequential IG method

Pull Request - State: closed - Opened by gsarti about 1 year ago
Labels: enhancement

#221 - Long prompt for decoder only model

Issue - State: closed - Opened by saxenarohit about 1 year ago - 4 comments
Labels: bug

#220 - Fix LIME and Occlusion outputs

Pull Request - State: closed - Opened by gsarti about 1 year ago

#219 - Support to from_pretrained args

Issue - State: closed - Opened by BiEchi about 1 year ago - 1 comment
Labels: question

#218 - On the interoperability with ferret

Issue - State: closed - Opened by romainf28 about 1 year ago - 3 comments
Labels: question

#217 - ALTI attribution method

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

#216 - Add model config

Pull Request - State: closed - Opened by gsarti about 1 year ago

#215 - Fix tokenization space cleanup

Pull Request - State: closed - Opened by gsarti over 1 year ago
Labels: bug

#214 - How can I anaylse integrated_gradients for Decoder-Only models?

Issue - State: closed - Opened by MaggieK410 over 1 year ago - 6 comments
Labels: question

#213 - Support `ContiguousSpanAggregation` when `attr_pos_start != 0`

Pull Request - State: closed - Opened by gsarti over 1 year ago
Labels: bug

#212 - `SubwordAggregator` challenging to use for non-SentencePiece models

Issue - State: closed - Opened by gsarti over 1 year ago
Labels: enhancement

#211 - Bump certifi from 2023.5.7 to 2023.7.22

Pull Request - State: closed - Opened by dependabot[bot] over 1 year ago - 2 comments
Labels: dependencies

#210 - fix `merge_attributions`

Pull Request - State: closed - Opened by DanielSc4 over 1 year ago - 5 comments
Labels: bug

#209 - Confusing call of `merge_attributions` method

Issue - State: closed - Opened by DanielSc4 over 1 year ago - 1 comment
Labels: bug

#208 - Visualize Attention Weights for a Decoder Only Model

Issue - State: closed - Opened by aflah02 over 1 year ago - 2 comments
Labels: question

#207 - Allow contrastive attribution with shorter contrastive targets

Pull Request - State: closed - Opened by gsarti over 1 year ago
Labels: enhancement

#206 - Add `top_p_size` step fn, `StepFunctionArgs` class

Pull Request - State: closed - Opened by gsarti over 1 year ago

#205 - Support `petals` distributed model classes

Pull Request - State: closed - Opened by gsarti over 1 year ago
Labels: enhancement

#204 - Fix attribution remapping for decoder-only models

Pull Request - State: closed - Opened by gsarti over 1 year ago

#203 - Add Saliency Cards to documentation

Issue - State: open - Opened by gsarti over 1 year ago
Labels: enhancement, help wanted, good first issue, user qol

#202 - Add `scores_precision` parameter to `FeatureAttributionOutput.save`

Issue - State: open - Opened by gsarti over 1 year ago
Labels: enhancement, help wanted, good first issue

#201 - Improved quickstart documentation

Pull Request - State: closed - Opened by gsarti over 1 year ago

#200 - Add ALTI+ implementation

Issue - State: open - Opened by gsarti over 1 year ago - 2 comments
Labels: enhancement

#199 - Remove forced seed in attribution

Pull Request - State: closed - Opened by gsarti over 1 year ago

#198 - Error of input text not matching decoded output of tokenizer

Issue - State: closed - Opened by DanielSc4 over 1 year ago
Labels: bug

#197 - Add support for PEFT models

Issue - State: open - Opened by DanielSc4 over 1 year ago - 1 comment
Labels: enhancement, to be investigated

#196 - Add example tutorial

Pull Request - State: closed - Opened by gsarti over 1 year ago

#194 - implementation of get_post_variable_assignment_hook

Issue - State: closed - Opened by martijnvanbeers over 1 year ago - 2 comments
Labels: question

#193 - Tokens diff view for contrastive attribution methods

Pull Request - State: closed - Opened by gsarti over 1 year ago

#192 - Fix `get_scores_dict` for duplicate tokens

Pull Request - State: closed - Opened by gsarti over 1 year ago
Labels: bug

#191 - `CUDA out of memory` for larger datasets during attribution

Issue - State: open - Opened by lsickert over 1 year ago - 9 comments
Labels: bug

#190 - GPT2 Integrated Gradients - empty input gives false results

Issue - State: closed - Opened by Victordmz over 1 year ago - 2 comments
Labels: bug, to be investigated

#189 - Limit the plots displayed through `.show()` function

Issue - State: open - Opened by lsickert over 1 year ago
Labels: enhancement, user qol