Ecosyste.ms: Issues

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

GitHub / google/neural-tangents issues and pull requests

#206 - Error with batching

Issue - State: closed - Opened by HaozhenZhao 3 months ago

#205 - Support python 3.12 and misc

Pull Request - State: open - Opened by romanngg 3 months ago - 1 comment

#204 - High Memory Usage on Infinite-Width NTK for GPU Only

Issue - State: open - Opened by deoliveirajoshua 4 months ago - 1 comment

#202 - Tag a release

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

#200 - Question regarding the cookbook

Issue - State: open - Opened by XinyiYS 9 months ago

#199 - Inefficient jacobian computation for embedding layers.

Issue - State: open - Opened by mohamad-amin 10 months ago - 1 comment
Labels: enhancement

#198 - NKT_mean output Nan, when the number of training sample is increased

Issue - State: open - Opened by zhangbububu 10 months ago - 3 comments
Labels: question

#196 - How to use batch to gradient_descent_mse_ensemble ?

Issue - State: closed - Opened by yanglebupt 12 months ago - 1 comment
Labels: question

#195 - stax.serial PSDness

Issue - State: closed - Opened by jaya-pret 12 months ago - 1 comment
Labels: question

#193 - NTK is not PD

Issue - State: closed - Opened by ShenLi00011 about 1 year ago

#192 - using stax.Cos(a=1.0, b=1.0, c=0.0) to get kernel from conv layer gives error

Issue - State: closed - Opened by ShenLi00011 about 1 year ago - 2 comments
Labels: question

#191 - Erf function goes beyond [-1,1]

Issue - State: open - Opened by bangxiangyong about 1 year ago - 2 comments
Labels: bug

#190 - pip install issues

Issue - State: closed - Opened by SamTov about 1 year ago - 2 comments
Labels: bug

#189 - How to compute the empirical after kernel?

Issue - State: open - Opened by VMS-6511 about 1 year ago - 1 comment
Labels: question

#188 - How to obtain aleatoric uncertainty?

Issue - State: open - Opened by bangxiangyong over 1 year ago - 2 comments
Labels: question

#187 - Colab notebooks issue

Issue - State: closed - Opened by juan190199 over 1 year ago - 2 comments
Labels: bug

#186 - eNTK implementation uses deprecated xla attribute

Issue - State: closed - Opened by sim0000 over 1 year ago - 2 comments
Labels: bug

#185 - Correct arxiv links in README.md

Pull Request - State: closed - Opened by mmhamdy over 1 year ago - 3 comments

#184 - Question regarding lr in Neural Tangents Cookbook

Issue - State: closed - Opened by alicedimo over 1 year ago

#182 - Question regarding OOM issues

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

#180 - Question: Relu Kernel Computation

Issue - State: closed - Opened by yCobanoglu over 1 year ago - 3 comments
Labels: question

#178 - The analytical output of GP can not fit the result of NNGP generated by the nt.predict.gp_inference

Issue - State: open - Opened by NaCL16 over 1 year ago - 1 comment
Labels: question

#176 - How to do Aggregate on a Graph whose nodes are all vectors

Issue - State: open - Opened by SchenbergZY over 1 year ago - 5 comments
Labels: question

#175 - Report of Abuse of misinformation in scientific research

Issue - State: closed - Opened by yhyu13 over 1 year ago

#174 - Are the parameters of the elementwise nonlinear layers trainable?

Issue - State: closed - Opened by rishiedupalli almost 2 years ago - 4 comments
Labels: question

#173 - AttributeError: module 'neural_tangents' has no attribute 'utils'

Issue - State: open - Opened by LeavesLei almost 2 years ago - 8 comments
Labels: bug

#172 - Defining Own Kernel Function in Neural Tangents

Issue - State: open - Opened by snehjp2 almost 2 years ago - 1 comment
Labels: question

#171 - Conflicting dependencies between Jax and neural-tangents

Issue - State: closed - Opened by snehjp2 almost 2 years ago - 6 comments
Labels: bug

#170 - The different output dimensions in examples

Issue - State: closed - Opened by cycyyy almost 2 years ago - 2 comments
Labels: question

#169 - Cases where NTK and NNGP are very different

Issue - State: open - Opened by YehonatanAvidan almost 2 years ago
Labels: question

#168 - Excessive memory consumption for deep networks

Issue - State: open - Opened by jglaser about 2 years ago - 2 comments
Labels: enhancement

#166 - Questions about memory consumption of infinitely wide NTK

Issue - State: open - Opened by jasonli0707 about 2 years ago - 6 comments
Labels: bug

#165 - Does neural-tangents work for custom layer?

Issue - State: open - Opened by Shuhul24 about 2 years ago - 1 comment
Labels: question

#164 - Draw Phase Diagram for CNTK

Issue - State: open - Opened by hhorace about 2 years ago - 1 comment
Labels: enhancement

#163 - Question: simple example poor performance, what am I doing wrong?

Issue - State: open - Opened by mfouesneau over 2 years ago - 5 comments
Labels: question

#162 - neural tangents vs flax

Issue - State: closed - Opened by po-oya over 2 years ago - 3 comments
Labels: question

#160 - Why is the covariance a 2D matrix?

Issue - State: open - Opened by LeonhardStorm over 2 years ago - 2 comments
Labels: question

#159 - Feature masks do not get reduced in the kernel

Issue - State: open - Opened by jglaser over 2 years ago - 2 comments
Labels: bug

#158 - Fix masking with more than one input feature

Pull Request - State: open - Opened by jglaser over 2 years ago - 1 comment

#157 - Use local device count for pmap in np.batch

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

#156 - Point JAX installs in colabs to the new CUDA releases page.

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

#155 - Neural tangents finite network gives different results compared with JAX

Issue - State: closed - Opened by NiuJ1ao over 2 years ago - 1 comment
Labels: question

#154 - Improve type annotations and documentation

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

#152 - predict.py raises an error

Issue - State: closed - Opened by zeinabsadeghian over 2 years ago - 7 comments
Labels: bug

#151 - How to analyze transductive node classification

Issue - State: open - Opened by joneswong over 2 years ago - 1 comment
Labels: question

#150 - Question about training model

Issue - State: closed - Opened by Firegreat123 over 2 years ago - 2 comments
Labels: question

#149 - `nt.empirical_ntk_fn` and `nt.linearize` do not compute gradients past the linear head in a ViT

Issue - State: closed - Opened by refitwhisper over 2 years ago - 2 comments
Labels: question

#147 - Memory profiling with JAX pprof

Issue - State: open - Opened by celidos over 2 years ago - 1 comment
Labels: question

#146 - Question: I can't release the memory in gpu during execute nt.empirical_ntk_fn

Issue - State: open - Opened by kkeevin123456 over 2 years ago - 4 comments
Labels: question

#144 - nt.linearize induces a CUDA_ERROR_OUT_OF_MEMORY error

Issue - State: open - Opened by RylanSchaeffer over 2 years ago - 7 comments
Labels: question

#143 - NTK for Inputs on Hypersphere

Issue - State: closed - Opened by lkskstlr almost 3 years ago - 2 comments
Labels: question

#142 - Created experimental folder and added all NTK sketching codes

Pull Request - State: open - Opened by insuhan almost 3 years ago - 6 comments

#141 - Internal change

Pull Request - State: open - Opened by romanngg almost 3 years ago - 1 comment

#140 - Add Pytype support

Pull Request - State: closed - Opened by romanngg almost 3 years ago - 1 comment

#139 - Imports not working

Issue - State: closed - Opened by alicedimo almost 3 years ago - 2 comments
Labels: bug

#138 - Strange result with a simple 2-layers NN

Issue - State: open - Opened by jecampagne almost 3 years ago - 3 comments
Labels: bug

#137 - Question: difference between kernel_fn from stax.serial and nt.empirical_ntk_fn(apply_fn)?

Issue - State: open - Opened by jecampagne almost 3 years ago - 2 comments
Labels: question

#135 - Question: input dimension ???

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

#134 - Non-PSD kernel encountered when computing Empirical NTK

Issue - State: open - Opened by mohamad-amin almost 3 years ago - 3 comments

#133 - Suggestion: Replace _scan in `batch.py` with `lax.scan`

Issue - State: open - Opened by mohamad-amin almost 3 years ago
Labels: enhancement

#132 - Question about Standard Parameterization

Issue - State: open - Opened by bwnjnOEI almost 3 years ago - 4 comments
Labels: question

#131 - Questions about wide networks

Issue - State: open - Opened by JinraeKim almost 3 years ago - 3 comments
Labels: question

#130 - Question about the choice of batch size in kernel computation

Issue - State: open - Opened by mohamad-amin about 3 years ago - 2 comments
Labels: enhancement

#129 - Mismatch between implementation and documentation

Issue - State: closed - Opened by mekty2012 about 3 years ago - 3 comments
Labels: bug

#128 - Gettting NaN for predict.gradient_descent_mse

Issue - State: closed - Opened by won-bae about 3 years ago - 5 comments
Labels: question

#127 - colab cookbook issue

Issue - State: closed - Opened by frank-ccc about 3 years ago - 3 comments
Labels: bug

#126 - Possible bug in calculation of the value of the function at time t

Issue - State: open - Opened by mohamad-amin about 3 years ago - 9 comments
Labels: question

#125 - Jax v0.2.21 compatibility

Issue - State: closed - Opened by PythonNut about 3 years ago - 3 comments
Labels: bug

#124 - Significant memory leak while using predict.gradient_descent

Issue - State: closed - Opened by won-bae about 3 years ago - 4 comments
Labels: bug

#123 - value_and_grad(kernel_fn) not equal to kernel_fn with standard parameterization

Issue - State: closed - Opened by PythonNut about 3 years ago - 8 comments
Labels: bug

#122 - Significant difference in empirical NTK for batched and non-batched versions

Issue - State: closed - Opened by gortizji about 3 years ago - 5 comments
Labels: bug

#121 - NotImplementedError: When I use stax.DotGeneral

Issue - State: open - Opened by kkeevin123456 over 3 years ago - 4 comments
Labels: question

#120 - Is there a way to compute the trace of the empirical NTK?

Issue - State: open - Opened by geajack over 3 years ago - 1 comment
Labels: question

#119 - Question: nt.stax vs flax (or other JAX-based NN libs)

Issue - State: open - Opened by vimalthilak over 3 years ago - 1 comment
Labels: question

#118 - Question: is this the correct way to combine stax.Aggregate and nt.predict?

Issue - State: closed - Opened by bestadcarry over 3 years ago - 5 comments
Labels: question

#117 - Softmax layers?

Issue - State: open - Opened by geajack over 3 years ago - 7 comments
Labels: question

#116 - How to sample functions from posterior ?

Issue - State: open - Opened by bangxiangyong over 3 years ago - 2 comments
Labels: question

#115 - about precomputed kernel

Issue - State: closed - Opened by sassan72 over 3 years ago - 1 comment
Labels: question

#114 - What's the prior distribution of NNGP and NTK?

Issue - State: open - Opened by Kangfei over 3 years ago - 3 comments
Labels: question

#113 - Question: kernel_fn implementation in Erf()

Issue - State: closed - Opened by ryuichi0704 over 3 years ago - 3 comments
Labels: question

#112 - A type error in predict.gradient_descent

Issue - State: closed - Opened by kim-hyunsu over 3 years ago - 7 comments
Labels: bug

#108 - wrong paper link

Pull Request - State: closed - Opened by themrzmaster over 3 years ago - 1 comment

#107 - neural_tangent.stax network does not produce same outputs as equivalent Keras network

Issue - State: closed - Opened by geajack over 3 years ago - 1 comment
Labels: question

#106 - Can this be used to compute the NTK for a finite-width neural network?

Issue - State: closed - Opened by geajack over 3 years ago - 8 comments
Labels: question

#101 - Is it possible to implement the ELU Activation?

Issue - State: closed - Opened by nikhilvyas almost 4 years ago - 1 comment
Labels: enhancement

#100 - High GPU memory during empirical NTK calculation

Issue - State: closed - Opened by chenwydj almost 4 years ago - 5 comments
Labels: question

#99 - Can we get kernel_fn from flax models?

Issue - State: closed - Opened by chenwydj almost 4 years ago - 3 comments
Labels: enhancement

#98 - increase batch size

Pull Request - State: closed - Opened by romanngg almost 4 years ago

#97 - Don't use functools.partial to fix autodoc

Pull Request - State: closed - Opened by romanngg almost 4 years ago

#96 - Don't use type-altering decorators to make autodoc run

Pull Request - State: closed - Opened by romanngg almost 4 years ago

#95 - Update README.md

Pull Request - State: closed - Opened by sschoenholz almost 4 years ago

#94 - Update README.md

Pull Request - State: closed - Opened by romanngg almost 4 years ago

#93 - ntkgp option in predict.gp_inference

Pull Request - State: closed - Opened by bobby-he almost 4 years ago - 2 comments

#92 - Using gp_inference to predict the posterior distribution - What is nngp_test_test?

Issue - State: closed - Opened by samuelkim314 almost 4 years ago - 6 comments
Labels: question

#89 - batch-decorated kernel function gives out-of-memory error

Issue - State: open - Opened by qixuanf almost 4 years ago - 1 comment
Labels: bug