stanfordnlp/pyvene issues and pull requests

#93 - [Major] Update with string access and code refactory (#83)

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

#92 - Model utils tests for GPT-2

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

#91 - Update datagenerators to support tokenizing for LMs

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

#90 - Update datagenerators to support tokenizing for LMs

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

#89 - [Minor] fix requirements.txt to have a more stable Colab experience (#86)

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

#88 - More scatter_neurons() tests

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

#87 - [Bug fix] fix backpack init

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

#86 - [P0] Upgrade tutorials to new API

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

#85 - [Minor] Support backpack and replicate

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

#84 - Add an extra test on scatter_neurons() testing case with no head.

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

#83 - [P1] Optionally remove the dependency of the config file

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

#82 - Add preliminary scatter() function tests

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

#81 - gather_neurons() unit tests

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

#80 - [Minor] Add Viz

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

#79 - [Minor] Update a few doc links

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

#78 - [Major] Update API namings and parameters

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

#77 - [P2] Sparse autoencoders

Issue - State: closed - Opened by aryamanarora 11 months ago - 3 comments

#76 - [P1] Speed up training of multiple DAS interventions with caching

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

#75 - [Bug Fix] Topological order scoring is not accurate

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

#74 - Re-grouped utils, cleaned up tests

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

#73 - Bug in BoundlessRotatedSpaceIntervention

Issue - State: closed - Opened by jiudingsun01 11 months ago - 1 comment

#72 - Merge main into dev/models

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

#71 - Merge Mistral model to dev/models

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

#70 - Model generation API simplified and cleanup tech debt on redundant variables

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

#69 - [Minor] update broadcast in generation

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

#68 - [P1] Tutorial of Inference-time Intervention

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

#67 - [P2] Mac chip MPS mode support

Issue - State: open - Opened by frankaging 11 months ago - 1 comment

#66 - Zen/causaltracingfull

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

#65 - [Minor] incidental changes to the tutorials for updated results

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

#64 - [P1] Adding tests for interventions and their util functions

Issue - State: closed - Opened by frankaging 11 months ago

#63 - [P1] Adding tests for functions in `modeling_utils.py`

Issue - State: closed - Opened by frankaging 11 months ago

#62 - [P1] Supporting Mistral with Unit Tests (#46)

Issue - State: closed - Opened by frankaging 11 months ago - 2 comments

#61 - [Minor] Update to full reproduce result fig1 the ROME paper

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

#60 - [Minor] Add in full causal tracing fig 1 reproduced results

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

#59 - [P1] Adding in constant source intervention support with new tests

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

#58 - [Minor] update test name

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

#57 - Adding Mistral into pyvene

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

#56 - [Minor] add simple location broadcast for easy interface (#52)

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

#55 - [Minor] Minor change to test with num of label

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

#54 - [P1] Multi-GPU model sharding with intervening evaluation and training

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

#53 - [P2] Quantized model support

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

#52 - [P1] Supports simple use case of intervention (fixed location intervention for all examples)

Issue - State: closed - Opened by frankaging 11 months ago

#51 - [Bug Fix] `MLPConfig` contains conflict field names with HF that cannot be set to other values when initializing

Issue - State: closed - Opened by frankaging 11 months ago

#50 - [Minor] allow dynamic boundary init for boundless das

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

#49 - Make self.embed_dim not None.

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

#48 - Remove default cache directory

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

#47 - update setup link, tutorial links

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

#46 - [P1] Support more huggingface (transformer-based) models

Issue - State: open - Opened by frankaging 11 months ago - 3 comments
Labels: good first issue

#45 - [P2] Having a simple tutorial on how to use intervenable model inside another torch module, or pipeline (e.g., HF pipeline)

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

#44 - Major update with updated class name; tutorial updates (#41)

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

#43 - Probing tutorial

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

#42 - Add SigmoidMaskRotatedSpaceIntervention

Pull Request - State: closed - Opened by jiudingsun01 11 months ago

#41 - [P0] Library Class Renaming Effort

Issue - State: closed - Opened by frankaging 11 months ago

#40 - [P1] Support base-only inputs with activation collection intervention

Issue - State: closed - Opened by frankaging 11 months ago

#39 - add individual sigmoid masks to Boundless DAS

Pull Request - State: closed - Opened by jiudingsun01 11 months ago

#38 - [P1] Unit tests for main class objects

Issue - State: closed - Opened by frankaging 11 months ago

#37 - [External] Save Interventions

Issue - State: closed - Opened by jiudingsun01 11 months ago - 1 comment

#36 - Add datagenerators and the IntroToDAS tutorial

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

#35 - [P2] Intervening through time recurrently without teacher-forcing

Issue - State: open - Opened by frankaging 11 months ago - 1 comment

#34 - [P2] Enable `use_fast` for head related interventions

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

#33 - [P0] enable `use_fast` option in the alignable to hyper boost training speed in case intervention locations (for position+subspace) are fixed in a batch

Issue - State: closed - Opened by frankaging 11 months ago - 2 comments

#32 - [P1] Add unit test for vanilla intervention at different positions and streams

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

#31 - Support BLIP model (visual question answering)

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

#30 - [P1] Streamlining trainable intervention artifacts saving and sharing

Issue - State: closed - Opened by frankaging 12 months ago

#29 - [P1] Initial support of stateful model architectures such as RNN

Issue - State: closed - Opened by frankaging 12 months ago

#28 - [External] Counterfactual sampling in price-tagging tutorial utils code

Issue - State: closed - Opened by dangng2004 12 months ago - 4 comments

#27 - add in group internvetion (#22) and code cleanupLocal dev

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

#26 - [P2] Probe training with interventions

Issue - State: closed - Opened by frankaging about 1 year ago - 2 comments

#25 - [P2] Support simple MLP layer for interventions

Issue - State: closed - Opened by frankaging about 1 year ago

#24 - [P2] Support interchange intervention training by unfreezing model weights with vanilla intervention

Issue - State: closed - Opened by frankaging about 1 year ago

#23 - [P2] Refactor utils in models to smaller files

Issue - State: closed - Opened by frankaging about 1 year ago

#22 - [P1] Introduce grouped interventions where each group is associated with the same source input

Issue - State: closed - Opened by frankaging about 1 year ago

#21 - [P2] Supports list of subspace indices instead of chunk partition

Issue - State: closed - Opened by frankaging about 1 year ago

#20 - [P2] Refactor the AlignableConfig to take in intervention not as type but actual instance

Issue - State: closed - Opened by frankaging about 1 year ago

#19 - [P1] Unit test with hand-crafted models

Issue - State: closed - Opened by frankaging about 1 year ago - 2 comments

#18 - [P1] Support multi-variable alignment with a single trainable basis intervention

Issue - State: closed - Opened by frankaging about 1 year ago - 1 comment

#17 - [P1] Adding support to intervene selected positions of selected heads

Issue - State: closed - Opened by frankaging about 1 year ago

#16 - Fix hook counters for non-generation calls.

Pull Request - State: closed - Opened by jhuang81 about 1 year ago

#15 - [P0] Adding more intervention support in generate mode

Issue - State: closed - Opened by frankaging about 1 year ago

#14 - Kwargs

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

#13 - Causal tracing example

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

#12 - Add Pythia (GPT_NeoX) support

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

#11 - [P2] Support argument name based intervention

Issue - State: open - Opened by frankaging about 1 year ago

#10 - Fix intervention dimension.

Pull Request - State: closed - Opened by jhuang81 about 1 year ago

#9 - Move bf16 conversion to rotation layer

Pull Request - State: closed - Opened by jhuang81 about 1 year ago

#8 - Making this a generic library for aligning model internals with customizable interventions

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

#7 - Contributions from dduan's cs224u project

Pull Request - State: closed - Opened by dduan97 over 1 year ago

#6 - Contributions from dduan's cs224u project

Pull Request - State: closed - Opened by dduan97 over 1 year ago

#5 - Error when running run_alignment.py: has no attribute when trying to save rotation matrix.

Issue - State: closed - Opened by rosafish over 1 year ago - 2 comments

#4 - Add "LlamaForCausalLM" architecture name

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

#3 - LlamaForCausalLM (case sensitive issue?)

Issue - State: closed - Opened by SungMinCho over 1 year ago - 2 comments

#2 - Icl

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

#1 - add in gpt2 alignable, separate trainer/metrics

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

GitHub / stanfordnlp/pyvene issues and pull requests