Mateen Ulhaq comments

Results 162 comments of


                                            Mateen Ulhaq

error when trying to use entropy bottleneck of compressai in my code

That says the module `compressai` is not available in Python's paths. Try installing CompressAI using the instructions in https://github.com/InterDigitalInc/CompressAI?tab=readme-ov-file#installation. In your case, this means creating a virtual environment, activating the...

How can I understand imposing an absolute value and a negative sign on a random variable before calculating the likelihood? Thank you very much for your answer! ! !

### Proof by symmetry Define: ```python def _standardized_cumulative(x): return 0.5 * torch.erfc(-x / sqrt(2)) ``` [Plot:](https://www.wolframalpha.com/input?i=0.5+*+erfc%28-x+%2F+sqrt%282%29%29) ![_standardized_cumulative](https://github.com/InterDigitalInc/CompressAI/assets/721196/8a94d814-68f5-44d1-bfa0-537181cc1b60) `_standardized_cumulative` has various symmetries. Notably: ```python _standardized_cumulative(x) == 1 - _standardized_cumulative(-x) ``` Thus:...

Does the number of bits estimated by the neural network same with that obtained by the actrual entropy encoder?

The `forward` function gives (what the model thinks is) the best probability distribution for encoding a given symbol. If the probability distribution used for encoding is different from the one...

Does the number of bits estimated by the neural network same with that obtained by the actrual entropy encoder?

Try the items mentioned in https://github.com/InterDigitalInc/CompressAI/issues/236#issuecomment-1594490483: > - Minimize aux_loss to 0 using the code from https://github.com/InterDigitalInc/CompressAI/pull/231, then re-run update and eval. [[Example](https://github.com/InterDigitalInc/CompressAI/issues/236#issuecomment-1602346172)] > - Try printing out the ranges...

(Duplicate) Add time_scale option

Requesting merge. Plugin recently featured here: https://www.reddit.com/r/neovim/comments/1gyt2ea/karb94neoscrollnvim_sphambasmearcursornvim_make/ --- EDIT: Looks like the PR is outdated. I made a little hack here: ```diff diff --git a/lua/neoscroll/scroll.lua b/lua/neoscroll/scroll.lua index fa1b06a..29b5b2d 100644 ---...

bug: [python] conditional `return` prematurely de-indents

Ah, you're right. I'd forgotten to set it up to not use treesitter's default (`indentexpr=nvim_treesitter#indent()`). I fixed it so it now loads correctly (`indentexpr=v:lua.require'nvim-yati.indent'.indentexpr()`). The current behavior is fine, though...

EntropyBottleneck with adjustable bin width

At a quick glance, this should work for training, though I wonder if the lossless entropy coder also needs adjustment. A simpler method might be to just rescale the outputs...

EntropyBottleneck with adjustable bin width

Apologies for the late response. It looks like that should work for the runtime codec. However, if you wanted to train it as well, you would also need to modify...

When I wanted to migrate that program to sequence signal data, I found that @register_model("mbt2018") The compress and decompress methods of class JointAutoregressiveHierarchicalPriors are very complex

The autoregression portion requires a loop of many steps at runtime. This is because the needed information for decoding only becomes available as the tensor is being decoded pixel-by-pixel from...

About the decoding speed of the ELIC model

The measured decoding latency may depend on the CPU, GPU, or other conditions. Their setup is mentioned here: > ## Supplementary Material > ### 2. Detailed experimental settings > >...