README
đź“Š
Building AI affordable tomorrow through distillation.
Abstract Powered Research
Team
non-profit
AI & ML interests
Curating systems for dynamic growth of AI mechanisms.
Recent Activity
AbstractPhilÂ
updated a
Space 1 day ago
AbstractPhilÂ
posted an update 1 day ago
Post
118
Today, I'll be determining the codebook capacity and utility potential for the larger batteries; Fresnel, Johanna, Grandmaster, Freckles, and Johanna-F variants, which should give a good indication of which models are capable of handling codebooks and which are more errant. The earlier all use SVD while the later do not. The differences are noted per and the behavior divergent.
I anticipate the D=16 will be more errant, and the final-state variants of those could very well be much more difficult or costly to inference as their axis bends are likely considerably harder to track. However, I'm confident that enough bounces will give the yield required so I'll set up some high-yield noise barrages to determine how much of them we can in fact extract from Johanna, and then set up similar barrages for images to map the internals of Fresnel and Grandmaster.
Grandmaster will be tricky, as it was an experimental Johanna-256 finetuned series meant to map sigma noised image inputs to recreate Fresnel behavioral output. Noised image goes in -> Fresnel-grade replication comes out in high res.
This allowed preliminary Dall-E Mini-esque VAE generation and will be explored further for the stereoscopic translation subsystem, to allow image generation in the unique format of diffusion that I was working out. I anticipate this system to be more than capable at making monstrosities, so I won't be posting TOO MANY prelims on this one, but the high-capacity potential of these noise makers are meaningfully powerful. Getting uniform codebooks in-place for these models will allow full transformer mapping downstream instead of just guess working the MSE piecemeal, which the earlier versions and variants were doing.
I'm straying from the CLS specifically for this series because CLS creates adjudicated pools of bias orbiting the INCORRECT orbiter some SVAE. The orbital target IS the soft-hand accumulated bias with the sphere-norm, so having a competitor isn't going to be a good option.
I anticipate the D=16 will be more errant, and the final-state variants of those could very well be much more difficult or costly to inference as their axis bends are likely considerably harder to track. However, I'm confident that enough bounces will give the yield required so I'll set up some high-yield noise barrages to determine how much of them we can in fact extract from Johanna, and then set up similar barrages for images to map the internals of Fresnel and Grandmaster.
Grandmaster will be tricky, as it was an experimental Johanna-256 finetuned series meant to map sigma noised image inputs to recreate Fresnel behavioral output. Noised image goes in -> Fresnel-grade replication comes out in high res.
This allowed preliminary Dall-E Mini-esque VAE generation and will be explored further for the stereoscopic translation subsystem, to allow image generation in the unique format of diffusion that I was working out. I anticipate this system to be more than capable at making monstrosities, so I won't be posting TOO MANY prelims on this one, but the high-capacity potential of these noise makers are meaningfully powerful. Getting uniform codebooks in-place for these models will allow full transformer mapping downstream instead of just guess working the MSE piecemeal, which the earlier versions and variants were doing.
I'm straying from the CLS specifically for this series because CLS creates adjudicated pools of bias orbiting the INCORRECT orbiter some SVAE. The orbital target IS the soft-hand accumulated bias with the sphere-norm, so having a competitor isn't going to be a good option.
- 3 replies
AbstractPhilÂ
posted an update 4 days ago
Post
103
My recent study in a nutshell shows a few important elements and everything else is technical.
* There are most definitely invariant architectural geometric states that persist and can be taught.
* They are not coincidental and the process works effectively on multiple data types and processes, not just noise. Noise is just fast to test with.
* Systems like SVD, Eigh, Conv, and the like - HELP align those systems for larger structures to produce amplified stability, but are not required for smaller structures, and the tests show even attention gets in the way at the smallest.
* Batched arrays, stacks, queues, and so on - all improve performance depending on the task.
* An SVAE battery is resolution agnostic, meaning with simple processing and logic you can scan space and record meshes fairly optimally to record large amounts of inference data.
* Batteries when trained on one specific task often can be directly used for other tasks once a codebook is fitted with the necessary data. Meaning a battery trained on gaussian noise can be fed imagenet snippets and downstream the MSE rates from the 64 battery array can be consumed for statistics aggregation to a fair degree of accuracy without actually training the array on images themselves.
* The battery codebook is a pointwise rigid map within the battery and can be used for pairwise learning when using the H2, H2a, and H2b batteries.
So this is, the evolved state of the geometric vocabulary in some ways, and a completely new and unexpected systemic development in others. They stack, you can reuse them, so small you can swap them at runtime with no time loss, they align rapidly, and downstream tasks can consume their information.
There are many untested avenues that I need to make a full writeup for because quite frankly it's messy currently and Claude is only making it more messy instead of cleaner.
* There are most definitely invariant architectural geometric states that persist and can be taught.
* They are not coincidental and the process works effectively on multiple data types and processes, not just noise. Noise is just fast to test with.
* Systems like SVD, Eigh, Conv, and the like - HELP align those systems for larger structures to produce amplified stability, but are not required for smaller structures, and the tests show even attention gets in the way at the smallest.
* Batched arrays, stacks, queues, and so on - all improve performance depending on the task.
* An SVAE battery is resolution agnostic, meaning with simple processing and logic you can scan space and record meshes fairly optimally to record large amounts of inference data.
* Batteries when trained on one specific task often can be directly used for other tasks once a codebook is fitted with the necessary data. Meaning a battery trained on gaussian noise can be fed imagenet snippets and downstream the MSE rates from the 64 battery array can be consumed for statistics aggregation to a fair degree of accuracy without actually training the array on images themselves.
* The battery codebook is a pointwise rigid map within the battery and can be used for pairwise learning when using the H2, H2a, and H2b batteries.
So this is, the evolved state of the geometric vocabulary in some ways, and a completely new and unexpected systemic development in others. They stack, you can reuse them, so small you can swap them at runtime with no time loss, they align rapidly, and downstream tasks can consume their information.
There are many untested avenues that I need to make a full writeup for because quite frankly it's messy currently and Claude is only making it more messy instead of cleaner.
- 2 replies
AbstractPhilÂ
posted an update 6 days ago
Post
75
Ever see a 1024x1024 3 channel a little over 1m param noise classifier? This is one. This is phase 1 of the omega experiments and it's successful on a very high-accuracy selectivity level through statistics aggregation and pooling via... a tiny MLP attached to the battery array.
SVAE don't care what resolution you use. They never had that concern, they are solvers that fly through solutions. Perfect for math solutions of many formats and many structures, exactly what I need for the next stages.
AbstractPhil/geolip-svae-h2-64
Currently the primary use case for tests is noise format identification. There are multiple experiments to go before a full nth classification system is ready, however as it stands the only stopping point is training batteries now. They mostly train within about 10 million samples of tiny data so they will fly out hundreds a day if I find purposes for them.
Also I trained too many gaussian-related batteries, so there's really only about 50-100 or so batteries useful in the 192 array I set up. There's really only 64 batteries trained total but there are multiple epochs involved.
Now that there is a 57k parameter variation that converges on 16 variants of random noise like Johanna and Freckles before, you ask this model questions differently. You check the MSE to train downstream models, so if your array isn't conclusively working it won't work just yet.
It's not perfect yet, but it's improving daily.
A bad battery in the mix can be replaced at runtime.
SVAE don't care what resolution you use. They never had that concern, they are solvers that fly through solutions. Perfect for math solutions of many formats and many structures, exactly what I need for the next stages.
AbstractPhil/geolip-svae-h2-64
Currently the primary use case for tests is noise format identification. There are multiple experiments to go before a full nth classification system is ready, however as it stands the only stopping point is training batteries now. They mostly train within about 10 million samples of tiny data so they will fly out hundreds a day if I find purposes for them.
Also I trained too many gaussian-related batteries, so there's really only about 50-100 or so batteries useful in the 192 array I set up. There's really only 64 batteries trained total but there are multiple epochs involved.
Now that there is a 57k parameter variation that converges on 16 variants of random noise like Johanna and Freckles before, you ask this model questions differently. You check the MSE to train downstream models, so if your array isn't conclusively working it won't work just yet.
It's not perfect yet, but it's improving daily.
A bad battery in the mix can be replaced at runtime.
==============================================================================
PHASE J VERDICT
==============================================================================
Subset: 18 batteries, 1,029,870 params (vs 10.9M for full array)
Resolution A (summary) B (attn-pool)
256 96.6% 93.1%
512 95.4% 92.0%
1024 95.4% 95.4%- 5 replies
AbstractPhilÂ
posted an update 15 days ago
Post
139
The geolip-svd-transformer is almost ready.
I've spent multiple days preparing the substructure, scaling, testing, and expanding the system. The conduit is meant to reorganize data. Just like the SVAE prototypes, they are meant to sort and organize, not compress and compact.
The organization is almost prepared and almost ready. The resulting structure will produce projection-capable geometric aligned memory, compacted and transformed into a utilizable token set. The remaining structural components are specifically SVD-related utilities, and each of those are utilizing the variant natures of how difficult, how dispersed, and so on each component is as it's learned over time.
The SVAE components were perfect for testing this playground. They appear to be larger when analyzed, however the representation of those are meant to represent huge vocabularies. Patch 16x16 expanded upward to 768 is meant to encapsulate the behavior of near-pi upscaled, condensed into a considerably simpler smaller form.
This model is behaving perfectly. It does not encode in the traditional sense, it analyzes and produces geometric opinions throughout it's structure. Each of them proved one after the other the model could not only learn, but it can perfectly reconstruct, and with that produce utility-driven expansion capacity directly.
Fresnel -> effective image analysis battery.
Johanna -> effective noise analysis
Grandmaster -> Johanna finetuned with sigma restoration using Fresnel's opinions.
Freckles -> massive analysis array for noise (4096 to 16k tks)
Geometric batteries.
Cayley rotation is meant to encapsulate that potential and expand it, allowing further differentiation down the chain of model structural behavioral events.
Suffice it to say, this is the geometric transformer's evolved state. These will exist as conduits throughout the models, the expanded behavioral attenuation units meant to provide geometric analysis internally within models for data-oriented CV alignment.
I've spent multiple days preparing the substructure, scaling, testing, and expanding the system. The conduit is meant to reorganize data. Just like the SVAE prototypes, they are meant to sort and organize, not compress and compact.
The organization is almost prepared and almost ready. The resulting structure will produce projection-capable geometric aligned memory, compacted and transformed into a utilizable token set. The remaining structural components are specifically SVD-related utilities, and each of those are utilizing the variant natures of how difficult, how dispersed, and so on each component is as it's learned over time.
The SVAE components were perfect for testing this playground. They appear to be larger when analyzed, however the representation of those are meant to represent huge vocabularies. Patch 16x16 expanded upward to 768 is meant to encapsulate the behavior of near-pi upscaled, condensed into a considerably simpler smaller form.
This model is behaving perfectly. It does not encode in the traditional sense, it analyzes and produces geometric opinions throughout it's structure. Each of them proved one after the other the model could not only learn, but it can perfectly reconstruct, and with that produce utility-driven expansion capacity directly.
Fresnel -> effective image analysis battery.
Johanna -> effective noise analysis
Grandmaster -> Johanna finetuned with sigma restoration using Fresnel's opinions.
Freckles -> massive analysis array for noise (4096 to 16k tks)
Geometric batteries.
Cayley rotation is meant to encapsulate that potential and expand it, allowing further differentiation down the chain of model structural behavioral events.
Suffice it to say, this is the geometric transformer's evolved state. These will exist as conduits throughout the models, the expanded behavioral attenuation units meant to provide geometric analysis internally within models for data-oriented CV alignment.
- 6 replies
AbstractPhilÂ
posted an update 22 days ago
Post
151
Say hello to surge resonance training. From random init, 1 epoch trained the 128x128 imagenet SVAE with test reconstruction over 99% accurate by epoch 1 to 99.9% accurate by epoch 5.
AbstractPhil/geolip-SVAE
Epoch 1 test recon error 0.0064
Epoch 2 test recon error 0.0022
Epoch 8 is now 0.000294
Epoch 12 is now 0.000206
Epoch 14 is now 0.000190
Epoch 18 is now 0.000187
Epoch 24 is now 0.000117
Epoch 30 landmark 0.000099
There are NO EXPERTS HERE. This is pure self learning. The model learns the entire behavioral set within 1 epoch to reconstruct imagenet's test set to a useful state. By epoch 12 a recon of 0.000202 recall is now measured. This means, 99.99% accuracy at RECONSTRUCTING the test set through the bottleneck, while simultaneously leaving a trail of centerwise extraction as rich or richer.
ONE epoch. Just one.
Took about 10 minutes to train an already converged epoch, and I set it up for 200 epochs. This model will not need 200 epochs. I'd be surprised if it needs 3.
What you're looking at here, is the emergence of surge resonance. The power of a single epoch when the geometric CV alignment hits the tuning fork of absolute resonant perfection and counterpointed with the concerto's dissonant harmonic response.
I give you, surge resonance.
The metrics will be ready by morning and I'll begin building utilities to figure out what went right and what went wrong.
This model is rewarded when it exists within the geometric spectrum while simultaneously dual punished when leaving. There is no benefit to stray, and the benefit to exist within prevents the model from leaving the validated CV band.
This allows the model to exist perfectly within the tuning fork resonance structure.
The model CONTINUES to refine, even when the CV drift has begun to drift away from home. The model has left home and is now seeking new proximity.
Upcoming training will be the 256x256, 512x512, 1024x1024, and larger if the model holds. Each will be named.
AbstractPhil/geolip-SVAE
Epoch 1 test recon error 0.0064
Epoch 2 test recon error 0.0022
Epoch 8 is now 0.000294
Epoch 12 is now 0.000206
Epoch 14 is now 0.000190
Epoch 18 is now 0.000187
Epoch 24 is now 0.000117
Epoch 30 landmark 0.000099
There are NO EXPERTS HERE. This is pure self learning. The model learns the entire behavioral set within 1 epoch to reconstruct imagenet's test set to a useful state. By epoch 12 a recon of 0.000202 recall is now measured. This means, 99.99% accuracy at RECONSTRUCTING the test set through the bottleneck, while simultaneously leaving a trail of centerwise extraction as rich or richer.
ONE epoch. Just one.
Took about 10 minutes to train an already converged epoch, and I set it up for 200 epochs. This model will not need 200 epochs. I'd be surprised if it needs 3.
What you're looking at here, is the emergence of surge resonance. The power of a single epoch when the geometric CV alignment hits the tuning fork of absolute resonant perfection and counterpointed with the concerto's dissonant harmonic response.
I give you, surge resonance.
The metrics will be ready by morning and I'll begin building utilities to figure out what went right and what went wrong.
This model is rewarded when it exists within the geometric spectrum while simultaneously dual punished when leaving. There is no benefit to stray, and the benefit to exist within prevents the model from leaving the validated CV band.
This allows the model to exist perfectly within the tuning fork resonance structure.
The model CONTINUES to refine, even when the CV drift has begun to drift away from home. The model has left home and is now seeking new proximity.
Upcoming training will be the 256x256, 512x512, 1024x1024, and larger if the model holds. Each will be named.
- 3 replies
AbstractPhilÂ
posted an update 25 days ago
Post
157
The geolip-transformer-v8 requires a fundamental rethinking of training a core structure.
I'll make this brief and to the point.
GEOLIP is an observer system at it's core. It watches, triangulates, and assists with correct answers.
Many experiments worked very well, many fell down and turned into a pile of broken circuits. The recent geometric-transformer being one of my biggest fumbles, still taught me many things about what I'm TRULY trying to accomplish here.
**Save money and lives**. Less hardware use for less need at inference. Train more calculations into a more reusable and accurate structure for near instant zero-shot or sequential inference.
In the process v8 unlocked a missing puzzle piece, EMA trajectory alignment compensation. I'm doing my best to build something that works.
The geolip distillation system is very powerful but requires much experimentation still.
* Genetic experiments were successful
* Data transfer experiments successful
* Analysis experiments successful - and expand large model accuracy
* Many distillation experiments were successful.
* The largest successes being the kernels, the distillation tools, and the geometric analysis systems.
With the good comes the bad, the faulty VITs, the simultaneous trains that fault, the internalized confusion that happens occasionally.
*** The observer NEEDS something to OBSERVE. If the observer observes the progressive development of point cloud structures, it learns how to observe THAT LEARNING PROCESS - drifting fault assessment.
*** In the process it DOES NOT learn how to improve the CE relations by embedding and compensating with anchored triangulation opinions.
BIGGEST CONCLUSION. Staged curriculum training.
These components must be DECOUPLED. One must be a compounding structural awareness beacon, the other must be an informationally aligned composition in a utilizable fashion.
This means stage-by-stage freeze/unfreeze processing. Independent task-oriented structural alignment.
I'll make this brief and to the point.
GEOLIP is an observer system at it's core. It watches, triangulates, and assists with correct answers.
Many experiments worked very well, many fell down and turned into a pile of broken circuits. The recent geometric-transformer being one of my biggest fumbles, still taught me many things about what I'm TRULY trying to accomplish here.
**Save money and lives**. Less hardware use for less need at inference. Train more calculations into a more reusable and accurate structure for near instant zero-shot or sequential inference.
In the process v8 unlocked a missing puzzle piece, EMA trajectory alignment compensation. I'm doing my best to build something that works.
The geolip distillation system is very powerful but requires much experimentation still.
* Genetic experiments were successful
* Data transfer experiments successful
* Analysis experiments successful - and expand large model accuracy
* Many distillation experiments were successful.
* The largest successes being the kernels, the distillation tools, and the geometric analysis systems.
With the good comes the bad, the faulty VITs, the simultaneous trains that fault, the internalized confusion that happens occasionally.
*** The observer NEEDS something to OBSERVE. If the observer observes the progressive development of point cloud structures, it learns how to observe THAT LEARNING PROCESS - drifting fault assessment.
*** In the process it DOES NOT learn how to improve the CE relations by embedding and compensating with anchored triangulation opinions.
BIGGEST CONCLUSION. Staged curriculum training.
These components must be DECOUPLED. One must be a compounding structural awareness beacon, the other must be an informationally aligned composition in a utilizable fashion.
This means stage-by-stage freeze/unfreeze processing. Independent task-oriented structural alignment.
- 2 replies
AbstractPhilÂ
posted an update 28 days ago
Post
174
My heavily engineered repo; https://github.com/AbstractEyes/pytorch-parallel-compiler has been directly integrated into the geofractal repo for v1.2, if you use the geofractal repo be sure to pull for potential performance increases.
The WideRouter will enable multiple core new features; the predominant two for our next experiment are as follows.
1. Directly integrated multi-opinion constellation structures. This will enable dynamic compiled expansions internally within the structure for huge performance gains.
2. Controllable stage-by-stage compilation. Each stage can be compiled or not. SVD being notoriously non-compiler friendly due to the linalg.egens, I will be addressing this particular function DIRECTLY soon. There will be no quarter for graph breaks.
If the WideRouter causes any major bugs or breaks with your code, bad calculations, incorrect deviated gradients, twisted or contorted dtype outputs, or any major compilation errors; please don't hesitate to open a pull request. Claude and I will abruptly solve any major issues.
Once everything is perfectly in-line and the graph matches, the transformer will have massive geometric performance boosts for huge structural basins with multiple layers of depth.
I will be addressing the linalg.eig+eigh directly in conjunction with multiple argsort functions that are causing huge performance dips. As well as addressing every single use of .item() that can present itself in the compiler's path.
After this, the ensemble topological transformer will be a-go. Which will enable quaternion, FlowMagnitude, FlowAlignment, FlowVelocity, FlowVelocityQuaternion, FlowVelocityOrbital, FlowVelocityPentachoron, and multiple other flow matching systems that will improve performance by dominating amounts inline with minimal overhead cost due to the precomputed geometric structure.
The ensembles will feature multiple simultaneous batched and segmented forms of learning meant to train the oscillation omega predictor "Beatrix".
The WideRouter will enable multiple core new features; the predominant two for our next experiment are as follows.
1. Directly integrated multi-opinion constellation structures. This will enable dynamic compiled expansions internally within the structure for huge performance gains.
2. Controllable stage-by-stage compilation. Each stage can be compiled or not. SVD being notoriously non-compiler friendly due to the linalg.egens, I will be addressing this particular function DIRECTLY soon. There will be no quarter for graph breaks.
If the WideRouter causes any major bugs or breaks with your code, bad calculations, incorrect deviated gradients, twisted or contorted dtype outputs, or any major compilation errors; please don't hesitate to open a pull request. Claude and I will abruptly solve any major issues.
Once everything is perfectly in-line and the graph matches, the transformer will have massive geometric performance boosts for huge structural basins with multiple layers of depth.
I will be addressing the linalg.eig+eigh directly in conjunction with multiple argsort functions that are causing huge performance dips. As well as addressing every single use of .item() that can present itself in the compiler's path.
After this, the ensemble topological transformer will be a-go. Which will enable quaternion, FlowMagnitude, FlowAlignment, FlowVelocity, FlowVelocityQuaternion, FlowVelocityOrbital, FlowVelocityPentachoron, and multiple other flow matching systems that will improve performance by dominating amounts inline with minimal overhead cost due to the precomputed geometric structure.
The ensembles will feature multiple simultaneous batched and segmented forms of learning meant to train the oscillation omega predictor "Beatrix".
- 5 replies
AbstractPhilÂ
posted an update about 1 month ago
Post
147
geolip-ryan-spearman, the first dedicated protein observation structure meant to expand the tooling of the observer modeling system and introducing additional introspective analysis to the equation for genetic mutation and abnormality.
AbstractPhil/geolip-esm2_t33_650M_UR50D
This model is based on edm2 33 650m from facebook, assessed with specific benchmarks to be around 50% accurate or so. I'll be improving those numbers by self distillation spectrum. The models will never see the validation data while unfrozen. The full spectrum of training tools are visible.
This is the first self-distillation observer prototype, and it works. Not as rapidly as I had hoped, but it most definitely works. The SVD was the missing piece of geometric solidity required to preserve full rotational behavioral control. The kernel made this possible for rapid iteration, and the first results are coming in.
This inherits much of the functionality from the CLIP_L and CLIP_G memory banks, while benefitting from the advanced research I performed while extracting CaptionBert 5x bert pooled captions for target points.
The primary driving point here is the sheer data size - and the important contributions of that data size to a full construct of geometric aligned data. There is a massive amount of very specific information, all curated, perfectly labeled, and organized in a way that can be... well not so easily accessed, but I did find a few ways in.
This data is highly accurate and forged through life for billions of years. This is what is there, this is what is expected, and I have the tooling - stage by stage, to not only develop a solution for the problem, but to fully contribute to an improved version with minimal hardware requirement for training.
This is real expectation and the results are pouring in hourly, this can improve models beyond a reasonable baseline while preserving the baseline's correctness.
AbstractPhil/geolip-esm2_t33_650M_UR50D
This model is based on edm2 33 650m from facebook, assessed with specific benchmarks to be around 50% accurate or so. I'll be improving those numbers by self distillation spectrum. The models will never see the validation data while unfrozen. The full spectrum of training tools are visible.
This is the first self-distillation observer prototype, and it works. Not as rapidly as I had hoped, but it most definitely works. The SVD was the missing piece of geometric solidity required to preserve full rotational behavioral control. The kernel made this possible for rapid iteration, and the first results are coming in.
This inherits much of the functionality from the CLIP_L and CLIP_G memory banks, while benefitting from the advanced research I performed while extracting CaptionBert 5x bert pooled captions for target points.
The primary driving point here is the sheer data size - and the important contributions of that data size to a full construct of geometric aligned data. There is a massive amount of very specific information, all curated, perfectly labeled, and organized in a way that can be... well not so easily accessed, but I did find a few ways in.
This data is highly accurate and forged through life for billions of years. This is what is there, this is what is expected, and I have the tooling - stage by stage, to not only develop a solution for the problem, but to fully contribute to an improved version with minimal hardware requirement for training.
This is real expectation and the results are pouring in hourly, this can improve models beyond a reasonable baseline while preserving the baseline's correctness.
- 3 replies
AbstractPhilÂ
posted an update about 1 month ago
Post
186
SVD + Scatterpoint2D is the official encoding structure of the geolip system as of the image encoding tests.
Both unattuned scatterpoint2d and triton-aligned SVD are a cut above the rest by a large margin.
https://github.com/kymatio/kymatio
https://huggingface.co/blog/AbstractPhil/svd-triton-kernel-optimization
AbstractPhil/svd-triton
AbstractPhil/geolip-hypersphere-experiments
Most kymatio tests were done on standard pytorch models that yielded higher accuracy than simple conv or transformers before overfitting, but not in every instance. Most common tested low-count cifar10 and cifar100 instances yielded more for less. Those are in the hypersphere-experiments notebooks and are viewable via huggingface tensorboard metrics.
The accuracy, retention, agreement, disagreement, and sheer capacity of the refined SVD kernel shows that full Procrustes alignment is not just crucial to distillation, but also entirely representable within encoders themselves as students.
This structure can representationally re-impose layer-by-layer which is what I tested, and this capture system can behave as a global regularization system, a selector, a behavioral adjudication structure, an encoding solidification unit, a trajectory systemic accumulator, an anchored differentiation unit, and about 30 other tests show - all of the above simultaneously.
The preliminary rapid-iteration capable kernel shows that not only can these behaviorally represent utility, but the noise-drift can be directly accounted for using systems like GELU, drop path, dropout, and other elements to learn to ignore that very noise that accumulates.
Attention is now officially deemed valid when utilized based on the tests and examples allowing preserved geometric structure after attention selection.
This encoding structure is substantially more durable than I can give credit for.
Surge is coming, exactly as predicted. Late I admit.
Both unattuned scatterpoint2d and triton-aligned SVD are a cut above the rest by a large margin.
https://github.com/kymatio/kymatio
https://huggingface.co/blog/AbstractPhil/svd-triton-kernel-optimization
AbstractPhil/svd-triton
AbstractPhil/geolip-hypersphere-experiments
Most kymatio tests were done on standard pytorch models that yielded higher accuracy than simple conv or transformers before overfitting, but not in every instance. Most common tested low-count cifar10 and cifar100 instances yielded more for less. Those are in the hypersphere-experiments notebooks and are viewable via huggingface tensorboard metrics.
The accuracy, retention, agreement, disagreement, and sheer capacity of the refined SVD kernel shows that full Procrustes alignment is not just crucial to distillation, but also entirely representable within encoders themselves as students.
This structure can representationally re-impose layer-by-layer which is what I tested, and this capture system can behave as a global regularization system, a selector, a behavioral adjudication structure, an encoding solidification unit, a trajectory systemic accumulator, an anchored differentiation unit, and about 30 other tests show - all of the above simultaneously.
The preliminary rapid-iteration capable kernel shows that not only can these behaviorally represent utility, but the noise-drift can be directly accounted for using systems like GELU, drop path, dropout, and other elements to learn to ignore that very noise that accumulates.
Attention is now officially deemed valid when utilized based on the tests and examples allowing preserved geometric structure after attention selection.
This encoding structure is substantially more durable than I can give credit for.
Surge is coming, exactly as predicted. Late I admit.
- 1 reply
AbstractPhilÂ
posted an update about 1 month ago
Post
229
I built an actionable todo based on current research, former research, and compounded a full spectrum of potentials for image encoding into pure geometric structures, hybrid geometric structures, partial geometric structures, and full spectrum analysis relational structures. Claude built the manifest based on our research after forming a full research spectrum to head into actionable directions.
AbstractPhil/geolip-hypersphere-experiments
I have to say before I continue, Claude managed to keep a large running manifest of our research, and with that list this was possible. Without that list, this would have been entirely devoid of purpose, and Claude would likely have not extracted the information in a utilizable state for this solution set.
I'll be running the full series of tests in conjunction with the constellation architecture. Either it survives, or something entirely new will form. Based on the results from these tests, the directions will evolve.
Either way, the most optimal and fastest methodologies for this system will be benchmarked and utilized as the primary use-cases. The slower and more obviously higher-resolution variations will be optimized as much as possible and solutions provided.
Lets do this right.
With that, the first experiment will be geolip-anchor-scattering and the structure will be based on the first in the list.
I will be updating posts based on benchmarks, landmarks, and new insights while the Bert data cooks.
AbstractPhil/geolip-hypersphere-experiments
I have to say before I continue, Claude managed to keep a large running manifest of our research, and with that list this was possible. Without that list, this would have been entirely devoid of purpose, and Claude would likely have not extracted the information in a utilizable state for this solution set.
I'll be running the full series of tests in conjunction with the constellation architecture. Either it survives, or something entirely new will form. Based on the results from these tests, the directions will evolve.
Either way, the most optimal and fastest methodologies for this system will be benchmarked and utilized as the primary use-cases. The slower and more obviously higher-resolution variations will be optimized as much as possible and solutions provided.
Lets do this right.
With that, the first experiment will be geolip-anchor-scattering and the structure will be based on the first in the list.
I will be updating posts based on benchmarks, landmarks, and new insights while the Bert data cooks.
- 4 replies
AbstractPhilÂ
posted an update about 1 month ago
Post
199
Clawd breadcrumb trail AbstractPhil/geolip-hypersphere-experiments
With this I'll begin forming Clawd interface utility with the geofractal router, which will allow Clawd to form agentic clouds of utility that can be datawise trained on the go with minimal hardware requirement. This is not ready yet, but it begins very soon.
The recent experiments have solved the alignment issue that crippled collectives and forced my hand into ensemble research instead.
With those recent experiments, the geofractal router will allow modularization structural capacity after some preliminary alignment adjustment and adjudication experimentation. This will enable the full collective differentiation through codified attribution.
In other words, adding and removing modular AI elements to contribute to aligned communication streams, all speaking the same language. This is an adjacent and more powerful result than the anticipated geovocab patchwork, and it yields substantially more effective agentic solutions than moving around a bulky embedding echo-chamber.
https://github.com/AbstractEyes/geofractal
Procrustes whitening orthogonality will allow adding and removing elements from geofractal routers given a small amount of prep data, while the anchors of expectation can stay as a snap-on element.
The most inquisitive and interested researchers can follow the trail to find all of the experiments. Web crawl it with clawd and you can probably create a unified rationality pretty quickly, but I doubt you'll like what you find. The journey was extensive and the failures outweighed the successes, but I did find the lightbulb.
The represented outcomes are either in my articles in huggingface, my civit articles, my github repos, my huggingface repos, or I forgot to upload them and they're in my colab notebook heap.
As most research yields, it is mostly failures. However, there are many successes in the mix. Many. If you need solutions, you can dredge the bog.
With this I'll begin forming Clawd interface utility with the geofractal router, which will allow Clawd to form agentic clouds of utility that can be datawise trained on the go with minimal hardware requirement. This is not ready yet, but it begins very soon.
The recent experiments have solved the alignment issue that crippled collectives and forced my hand into ensemble research instead.
With those recent experiments, the geofractal router will allow modularization structural capacity after some preliminary alignment adjustment and adjudication experimentation. This will enable the full collective differentiation through codified attribution.
In other words, adding and removing modular AI elements to contribute to aligned communication streams, all speaking the same language. This is an adjacent and more powerful result than the anticipated geovocab patchwork, and it yields substantially more effective agentic solutions than moving around a bulky embedding echo-chamber.
https://github.com/AbstractEyes/geofractal
Procrustes whitening orthogonality will allow adding and removing elements from geofractal routers given a small amount of prep data, while the anchors of expectation can stay as a snap-on element.
The most inquisitive and interested researchers can follow the trail to find all of the experiments. Web crawl it with clawd and you can probably create a unified rationality pretty quickly, but I doubt you'll like what you find. The journey was extensive and the failures outweighed the successes, but I did find the lightbulb.
The represented outcomes are either in my articles in huggingface, my civit articles, my github repos, my huggingface repos, or I forgot to upload them and they're in my colab notebook heap.
As most research yields, it is mostly failures. However, there are many successes in the mix. Many. If you need solutions, you can dredge the bog.
- 1 reply
AbstractPhilÂ
posted an update about 2 months ago
Post
155
geolip-vit-x34 - 34 expert vit. I can't train an extended version of 34 vits, but I can definitely run some experiments and make some starter weights with an anchor. That would yield a substantial amount of data.
AbstractPhil/bulk-coco-features
This... is going to be a odd one to describe. Based on the research with Bert, creating a uniformed patchwork using a multitude of vit composites will be very achievable. It shouldn't be soup, which is really hard to explain, but by creating a second geometric anchor, the system will align in a way that I could never predict without many more model analysis and must test. I simply didn't test all these vits for geometry, so this will be the test.
This is essentially 34 directly extracted views of coco, which is already prepared feature data. With this data, we have 34 experts that can distill into a single unified vit. I'm hesitant to even call this distillation anymore, it's more interpolative data alignment, and it's absurdly retentive.
ADDITIONALLY, we can anchor to frozen geolip-bert and create cross-contrast between the anchors for a learned anchor median, which will allow further integrations directly into the geometric core.
This will require a few overlapping internal mechanisms to guarantee vit differentiation, however I believe the full unified patchwork will be... different from what is currently known as a vit.
geolip-bert-vit will likely be cooking within the month. The alignment statistics say it will be... 100% accurate to the specifications.
I CAN prepare 34 vits worth of imagenet, but I would need probably 34 vits worth of laion aesthetics, which is substantially more than I currently have. In the process I would need to ensure everything isn't corrupt, and the captions are correctly synthesized in our expert student bert with the correct anchoring rotation.
Probably 3 vits is enough for the full version prototype, 34 vits for the bulk experiment.
AbstractPhil/bulk-coco-features
This... is going to be a odd one to describe. Based on the research with Bert, creating a uniformed patchwork using a multitude of vit composites will be very achievable. It shouldn't be soup, which is really hard to explain, but by creating a second geometric anchor, the system will align in a way that I could never predict without many more model analysis and must test. I simply didn't test all these vits for geometry, so this will be the test.
This is essentially 34 directly extracted views of coco, which is already prepared feature data. With this data, we have 34 experts that can distill into a single unified vit. I'm hesitant to even call this distillation anymore, it's more interpolative data alignment, and it's absurdly retentive.
ADDITIONALLY, we can anchor to frozen geolip-bert and create cross-contrast between the anchors for a learned anchor median, which will allow further integrations directly into the geometric core.
This will require a few overlapping internal mechanisms to guarantee vit differentiation, however I believe the full unified patchwork will be... different from what is currently known as a vit.
geolip-bert-vit will likely be cooking within the month. The alignment statistics say it will be... 100% accurate to the specifications.
I CAN prepare 34 vits worth of imagenet, but I would need probably 34 vits worth of laion aesthetics, which is substantially more than I currently have. In the process I would need to ensure everything isn't corrupt, and the captions are correctly synthesized in our expert student bert with the correct anchoring rotation.
Probably 3 vits is enough for the full version prototype, 34 vits for the bulk experiment.
- 3 replies
AbstractPhilÂ
posted an update about 2 months ago
Post
177
geolip-captionbert-8192
This bert is currently being distilled using 5 bert teachers using the conceptual captions dataset. The recall accuracy is based on the whitened procrustes alignment, and the losses reflect keeping that rotation aligned correctly.
The expectation from the smaller prototypes show this model will align to 100% accuracy recall based on the most optimal opinions based on the correct answer, aligning specifically to the correct answers in conjunction with all the geometric losses.
No joke, this may be the smallest, least computation, most accurate, and fastest bert I've trained thus far - and it will be based entirely on five teachers simultaneously feeding opinions through a relay hub.
This bert is currently being distilled using 5 bert teachers using the conceptual captions dataset. The recall accuracy is based on the whitened procrustes alignment, and the losses reflect keeping that rotation aligned correctly.
The expectation from the smaller prototypes show this model will align to 100% accuracy recall based on the most optimal opinions based on the correct answer, aligning specifically to the correct answers in conjunction with all the geometric losses.
No joke, this may be the smallest, least computation, most accurate, and fastest bert I've trained thus far - and it will be based entirely on five teachers simultaneously feeding opinions through a relay hub.
- 13 replies
AbstractPhilÂ
posted an update about 2 months ago
Post
146
I'll attempt to expand the geolip-clip to full sequence context window to encompass sequential learning.
AbstractPhil/geolip-clip-vit-large-patch14-ctx576
The memory pod is specifically meant to tune everything based on final state pooling, which is fine if you aren't trying to actually use sequential utility.
HOWEVER, there are many elemental biases that present themselves if attempting to USE the standard sequence of 77 in conjunction with this final pooled state. Even though the standard 77 is predominantly noise past token 10 it still houses considerable amounts of information in terms of utility, so this should be handled carefully. Zero-shot structures are a tricky structure to analyze, especially structures based on attention mechanisms instead of true sequential accumulation. I've noticed I need to watch them for quite a while before the real bugs show up.
As it stands the token pool is essentially [B, 7+8, 768] for pools. This contains a robust and highly complex representation of useful accumulated bidirectional attention data, so it's quite powerful.
I'll build a few prototypes and tap into some papers. I'll either come up with something or a reason why I didn't. The end result will either produce an anchor bank set of tokens [B, 15, 768] for pooling, or [B, 15, 77, 768] ideally - which should expand the sequence of the clip to 1,155 if successful. That doesn't necessarily mean this sequence will be more useful than the [b, 15, 768], but it will be representationally valid to the context window expansion.
I wouldn't hold out for a single full-sequence option in a single day, that's a lot of moving parts to analyze, not to mention highly impractical to train with. A smaller dose of this information would be necessary for rapid prototyping so it'll likely be packaged as such.
Well I spoke too soon. It's ready to play with.
AbstractPhil/geolip-clip-vit-large-patch14-ctx576-seq77
AbstractPhil/geolip-clip-vit-large-patch14-ctx576
The memory pod is specifically meant to tune everything based on final state pooling, which is fine if you aren't trying to actually use sequential utility.
HOWEVER, there are many elemental biases that present themselves if attempting to USE the standard sequence of 77 in conjunction with this final pooled state. Even though the standard 77 is predominantly noise past token 10 it still houses considerable amounts of information in terms of utility, so this should be handled carefully. Zero-shot structures are a tricky structure to analyze, especially structures based on attention mechanisms instead of true sequential accumulation. I've noticed I need to watch them for quite a while before the real bugs show up.
As it stands the token pool is essentially [B, 7+8, 768] for pools. This contains a robust and highly complex representation of useful accumulated bidirectional attention data, so it's quite powerful.
I'll build a few prototypes and tap into some papers. I'll either come up with something or a reason why I didn't. The end result will either produce an anchor bank set of tokens [B, 15, 768] for pooling, or [B, 15, 77, 768] ideally - which should expand the sequence of the clip to 1,155 if successful. That doesn't necessarily mean this sequence will be more useful than the [b, 15, 768], but it will be representationally valid to the context window expansion.
I wouldn't hold out for a single full-sequence option in a single day, that's a lot of moving parts to analyze, not to mention highly impractical to train with. A smaller dose of this information would be necessary for rapid prototyping so it'll likely be packaged as such.
Well I spoke too soon. It's ready to play with.
AbstractPhil/geolip-clip-vit-large-patch14-ctx576-seq77
- 4 replies
AbstractPhilÂ
posted an update about 2 months ago
Post
275
geolip-bertenstein-v1 - 5 experts chosen. A collective of shared transformer aligned experts, not a mixture of experts. Similar to a MOE, but not quite. This first prototype won't have the full mailing projection relay system afforded by the geofractal router, but it will definitely be a solid prototype.
It is not production ready yet, there needs to be a few upstream and downstream tools meant to consume and process the outputs to create useful representations.
This model will be able to text respond, use whisper, see with dinolip, code with codebert, and process proteins using esm2_t33_650m_ur50.
Our experts for the prototype are;
google-bert/bert-large-uncased
facebook/dinov2-large
microsoft/codebert-base
openai/whisper-large-v3
facebook/esm2_t33_650M_UR50
Not the smartest text model, but more than enough for this preliminary use case test setup. Text is predominantly meant to align and orient downward function, the entire machine is meant to be operated unilaterally as a collective, or independently through individual pairs requests via special token access.
This model will be capable of substantial power and feats as a prototype. It will be capable of seeing and processing differential equations utilizing dinov2 and esm2 data simultaneously, which can be used for downstream analysis - and I WILL use that data to create a more powerful connection between dinov2 tokens, protein tokens, video tokens, code tokens, and audio tokens.
This is the FIRST prototype of this case, and I will introduce video, genetics, shape analysis, pattern recognition processing, and a much more powerful and reusable text model.
The tests show the models can have differential communication through the geolip transformers after procrustes pairwise analysis and pentachoron CV protective measures.
Whitening procrustes for precalculation and center-aligning allows for a faster convergence, so that should help too.
It is not production ready yet, there needs to be a few upstream and downstream tools meant to consume and process the outputs to create useful representations.
This model will be able to text respond, use whisper, see with dinolip, code with codebert, and process proteins using esm2_t33_650m_ur50.
Our experts for the prototype are;
google-bert/bert-large-uncased
facebook/dinov2-large
microsoft/codebert-base
openai/whisper-large-v3
facebook/esm2_t33_650M_UR50
Not the smartest text model, but more than enough for this preliminary use case test setup. Text is predominantly meant to align and orient downward function, the entire machine is meant to be operated unilaterally as a collective, or independently through individual pairs requests via special token access.
This model will be capable of substantial power and feats as a prototype. It will be capable of seeing and processing differential equations utilizing dinov2 and esm2 data simultaneously, which can be used for downstream analysis - and I WILL use that data to create a more powerful connection between dinov2 tokens, protein tokens, video tokens, code tokens, and audio tokens.
This is the FIRST prototype of this case, and I will introduce video, genetics, shape analysis, pattern recognition processing, and a much more powerful and reusable text model.
The tests show the models can have differential communication through the geolip transformers after procrustes pairwise analysis and pentachoron CV protective measures.
Whitening procrustes for precalculation and center-aligning allows for a faster convergence, so that should help too.
- 2 replies
AbstractPhilÂ
posted an update about 2 months ago
Post
1985
I've... done it. This, with experts, achieves near 100% R1 retrieval accuracy on an adjacent - unseen by the fusion transformer - dataset with around 40k steps from the seen dataset. This means the language of the models are at least tested fused within the constraints, not just projected or estimated.
AbstractPhil/geolip-procrustes
I encourage EVERYONE who is curious to check my work. Check it, double check it, and triple check it.
These were aligned using COCO and then validated with Flickr. Entirely different datasets. The experts arbitrated and the alignment yielded the correct answers. Preliminary tests show that with almost no alignment requirement, the models can reach 100% R1 retrieval accuracy.
Not to be confused with validation accuracy for a classification model or a text encoder's text response, this allows multispectral communication between entirely different models for direct downstream consumption with almost no training for the chosen models.
I have a working procrustes experiment that learns adjacent manifolds within a reasonable spectrum and the speed is... well, 1 epoch with COCO using Bert-Large and DinoV2 that allows the models to align nearly perfectly. For some scales in the experiment it shows that the 3 set epochs aren't quite enough to align R1 to highest, while many align nearly immediately.
These two were an obvious pair to pick, 60% similarity and >90% spectral similarity.
The trainer transfers layers, learns embeddings, and more - all by sticking strictly to geometric boundaries and procrustes informational accumulation within a modulation model's constraints.
I have many experiments to run.
AbstractPhil/geolip-procrustes
I encourage EVERYONE who is curious to check my work. Check it, double check it, and triple check it.
These were aligned using COCO and then validated with Flickr. Entirely different datasets. The experts arbitrated and the alignment yielded the correct answers. Preliminary tests show that with almost no alignment requirement, the models can reach 100% R1 retrieval accuracy.
Not to be confused with validation accuracy for a classification model or a text encoder's text response, this allows multispectral communication between entirely different models for direct downstream consumption with almost no training for the chosen models.
I have a working procrustes experiment that learns adjacent manifolds within a reasonable spectrum and the speed is... well, 1 epoch with COCO using Bert-Large and DinoV2 that allows the models to align nearly perfectly. For some scales in the experiment it shows that the 3 set epochs aren't quite enough to align R1 to highest, while many align nearly immediately.
These two were an obvious pair to pick, 60% similarity and >90% spectral similarity.
The trainer transfers layers, learns embeddings, and more - all by sticking strictly to geometric boundaries and procrustes informational accumulation within a modulation model's constraints.
I have many experiments to run.
- 1 reply
AbstractPhilÂ
posted an update about 2 months ago
Post
274
The small projection-based approximator model for the geolip patchwork did not breach a certain level of accuracy as required by my specifications, so I've defaulted to harvesting direct geometric information from AI models until I get the comparative bounds required for a useful topology.
I must sincerely apologize for not solving this problem quickly.
This will take time. Without the approximator it's going to be considerably slower, but this model I begin training will be providing the approximations in a different way over time. As iterations progress, the system will conform to a huge array of geometric potentials and be capable at predicting those, but it will not be as powerful as the full patchmaker up front, and it will be slow training.
If I can get my hands on a cluster of A100's or H100's for a measure I'll make a post immediately, until then I must default to the slower process.
I really banked that the smaller version would have worked, but it simply couldn't hold complex topological shape without the correct boundaries being learnable AND endure entropic decay simultaneously. The only way to have a predominant shot at a full geometric shared language, is to make those boundaries learnable in the full spectrum of potentials, or at least more than I have placed on it.
I'll be refining my process in the coming days further, and I do apologize for pre-emptively announcing a potential that I have yet to fully explore.
There will be a full upgraded 38 shape geolip patchwork trained asap to fully encompass the Flux 1 AE spectrum, and another trained for SD15, SDXL, and Flux 2's VAE as well. These will accommodate DIRECT complex geometric patchwork learning, but not to the scale as promised yet. Autoregression is a complex mistress as many of you know, and I will be spending a great deal of time and compute analyzing all of the information required to build a uniformly useful and powerful autoregression patchwork to utilize as invariance to teaching.
I must sincerely apologize for not solving this problem quickly.
This will take time. Without the approximator it's going to be considerably slower, but this model I begin training will be providing the approximations in a different way over time. As iterations progress, the system will conform to a huge array of geometric potentials and be capable at predicting those, but it will not be as powerful as the full patchmaker up front, and it will be slow training.
If I can get my hands on a cluster of A100's or H100's for a measure I'll make a post immediately, until then I must default to the slower process.
I really banked that the smaller version would have worked, but it simply couldn't hold complex topological shape without the correct boundaries being learnable AND endure entropic decay simultaneously. The only way to have a predominant shot at a full geometric shared language, is to make those boundaries learnable in the full spectrum of potentials, or at least more than I have placed on it.
I'll be refining my process in the coming days further, and I do apologize for pre-emptively announcing a potential that I have yet to fully explore.
There will be a full upgraded 38 shape geolip patchwork trained asap to fully encompass the Flux 1 AE spectrum, and another trained for SD15, SDXL, and Flux 2's VAE as well. These will accommodate DIRECT complex geometric patchwork learning, but not to the scale as promised yet. Autoregression is a complex mistress as many of you know, and I will be spending a great deal of time and compute analyzing all of the information required to build a uniformly useful and powerful autoregression patchwork to utilize as invariance to teaching.
- 2 replies
AbstractPhilÂ
posted an update 2 months ago
Post
1598
GLIP - Geometric Linear Interpolative Patchwork aka geolip.
https://github.com/AbstractEyes/glip-autoencoder
To tinker with the topology directly you can play with it here, though I admit it's imperfect in this form - it's quite the tinker toy to see the effects of patching.
https://claude.ai/public/artifacts/697287e4-fa18-4753-8b57-904d5e2022ed
This is the repo that will contain the next experimental stage, which is based entirely on the research and structural boundaries applied by said research. It'll be a little rigid while I get Claude set up.
In order to directly train these layered topological response patchworks you must install and use the geovocab2, geofractal, and wide_compiler repos.
This is due to the wide_compiler's wide_linear high-speed efficiency for ensemble processing, the geovocab2 factory structure with multiple formulas including highly efficient designs meant for kernel compilation, and a series of reusable utilities in geofractal including some of the more complex losses and difficult to optimally tune gate structures surrounding them.
Many of the underlying formulas are outlined here;
AbstractPhil/geometric-experiment-history
Utilization and training USING the pretrained or untrained geolip patchwork will be as simple as loading the model in pytorch and will not require external dependencies of the geolip package, numpy, or pytorch depending on the task. It will come packaged with recommended losses but I encourage experimentation because I simply cannot cover all spectrums.
More details to come as development progresses. The system is coming together and the state of the utilizable autoencoder will be ready within a couple weeks. The entire system is built for convenience and reusability, so the structure will be built similarly to autoencoder systems that currently exist, with a few tweaks here and there for important elements - so the interface will be familiar to those who use it.
https://github.com/AbstractEyes/glip-autoencoder
To tinker with the topology directly you can play with it here, though I admit it's imperfect in this form - it's quite the tinker toy to see the effects of patching.
https://claude.ai/public/artifacts/697287e4-fa18-4753-8b57-904d5e2022ed
This is the repo that will contain the next experimental stage, which is based entirely on the research and structural boundaries applied by said research. It'll be a little rigid while I get Claude set up.
In order to directly train these layered topological response patchworks you must install and use the geovocab2, geofractal, and wide_compiler repos.
This is due to the wide_compiler's wide_linear high-speed efficiency for ensemble processing, the geovocab2 factory structure with multiple formulas including highly efficient designs meant for kernel compilation, and a series of reusable utilities in geofractal including some of the more complex losses and difficult to optimally tune gate structures surrounding them.
Many of the underlying formulas are outlined here;
AbstractPhil/geometric-experiment-history
Utilization and training USING the pretrained or untrained geolip patchwork will be as simple as loading the model in pytorch and will not require external dependencies of the geolip package, numpy, or pytorch depending on the task. It will come packaged with recommended losses but I encourage experimentation because I simply cannot cover all spectrums.
More details to come as development progresses. The system is coming together and the state of the utilizable autoencoder will be ready within a couple weeks. The entire system is built for convenience and reusability, so the structure will be built similarly to autoencoder systems that currently exist, with a few tweaks here and there for important elements - so the interface will be familiar to those who use it.
- 6 replies
AbstractPhilÂ
posted an update 2 months ago
Post
1344
The Rosetta Stone geometric vocabulary and the ramping up capacity.
What makes this particular invariant special, is the existence within all structures I've tested so far. I had Claude write up the direct article based on what we built together, but I've tested it on many substructures. This is flawed, and I have a series of answers to making it more accurate.
First a reconstruction from the ground up. This means each shape is specifically built upward from the substructure to the point of inductive deviance. This will be less quick at first and then build speed as I optimize like the last system did.
The "saddle" problem; the system detected saddles because there wasn't enough deviance in the shapes to attenuate to more cardinality and more aligned substructures. The blobs were around 30-40% of the overall patches, which interpolated into the others produced a fair approximation.
It MOST DEFINITELY did see those shapes in their voxel complexity. This is real.
https://claude.ai/public/artifacts/bf1256c7-726d-4943-88ad-d6addb263b8b
You can play with a public claude artifact dedicated to viewing the current shape spectrum - and with that know exactly why it's flawed.
The flawed and repetitive shapes. I rapid prototyped and there are multiple redundant shapes that simply don't classify well or at all. Not to mention the rotation simply doesn't help much of the time, or doesn't exist with many shapes. This will be rectified in the next variation.
Projecting to shared latent space as a catalyst to allow growing subjective geoflow matched step variance, rather than simply direct classification. This will theoretically allow for full channel-to-channel invariant features to be mapped from structure to structure, and the very formula that encapsulated them to be directly baked into the math rather than classified as a substructure analysis.
There are many challenges between here and there, so stay tuned my friends as I plot the geometric language of pretrained AI.
What makes this particular invariant special, is the existence within all structures I've tested so far. I had Claude write up the direct article based on what we built together, but I've tested it on many substructures. This is flawed, and I have a series of answers to making it more accurate.
First a reconstruction from the ground up. This means each shape is specifically built upward from the substructure to the point of inductive deviance. This will be less quick at first and then build speed as I optimize like the last system did.
The "saddle" problem; the system detected saddles because there wasn't enough deviance in the shapes to attenuate to more cardinality and more aligned substructures. The blobs were around 30-40% of the overall patches, which interpolated into the others produced a fair approximation.
It MOST DEFINITELY did see those shapes in their voxel complexity. This is real.
https://claude.ai/public/artifacts/bf1256c7-726d-4943-88ad-d6addb263b8b
You can play with a public claude artifact dedicated to viewing the current shape spectrum - and with that know exactly why it's flawed.
The flawed and repetitive shapes. I rapid prototyped and there are multiple redundant shapes that simply don't classify well or at all. Not to mention the rotation simply doesn't help much of the time, or doesn't exist with many shapes. This will be rectified in the next variation.
Projecting to shared latent space as a catalyst to allow growing subjective geoflow matched step variance, rather than simply direct classification. This will theoretically allow for full channel-to-channel invariant features to be mapped from structure to structure, and the very formula that encapsulated them to be directly baked into the math rather than classified as a substructure analysis.
There are many challenges between here and there, so stay tuned my friends as I plot the geometric language of pretrained AI.
- 2 replies