Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

VPRTempo v1.1.0 #10

Merged
merged 73 commits into from
Dec 1, 2023
Merged

VPRTempo v1.1.0 #10

merged 73 commits into from
Dec 1, 2023

Conversation

AdamDHines
Copy link
Collaborator

Ready to merge to QVPR repo the newest version of VPRTempo.

…ocesses everthing and loads data through torch. Fixed up the weight additions, each layer with a True flag for weight additions is responsible for storing excitatory and inhibitory weights
…roblems, benched the system for speed compared to vanilla VPRTempo and is comparable when comparing image import times
…lues, finalized the output layer spike force training, adapted single training function to handle training of all layers
…-training, need to configure the test input tiling across modules
…ight update is adding is very high - investigate further
…hat bmm not supported for quantized tensors so need to explore a bit further
…with nn.Linear - removed modularity for now. Working on QAT, can train but cannot evaluate due to x and x_input tensors not quantized.
…rbage out. Looking like some x and x_input values are adding gradients and getting big over time, investigate further.
…quantized spikes is set to 1 so essentially all temporal information lost
@AdamDHines AdamDHines merged commit dbf31d5 into QVPR:main Dec 1, 2023
AdamDHines added a commit that referenced this pull request Jul 13, 2024
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

Successfully merging this pull request may close these issues.

1 participant