AbstractPhil/anima-90k-cache
Updated
AbstractPhila PRO
AbstractPhil
AI & ML interests
datasets, research papers, experimentation, vision, classification, text encoders, tokenization, llms, diffusion, distillation, and more.
Recent Activity
updated a dataset about 4 hours ago
AbstractPhil/anima-90k-cache published a dataset about 7 hours ago
AbstractPhil/anima-90k-cache published an article about 9 hours ago
Subject Bucketing: Teaching a Diffusion Model New Prompt Languages Without ForgettingOrganizations
updated a
dataset about 4 hours ago
published a
dataset about 7 hours ago
published an article about 9 hours ago
Article
Subject Bucketing: Teaching a Diffusion Model New Prompt Languages Without Forgetting
AbstractPhil
• posted an update about 9 hours ago
Post
25
Anima - Brent JSON (PREVIEW) - Subject Bucketing
Full article available https://huggingface.co/blog/AbstractPhil/subject-bucketing.
There is additionally a civit model release as well.
https://civitai.com/models/2730503/anima-jsonenglish
AbstractPhil/anima-prelim-1k-r64
The JSON multi-prompt diffusion model prototype using Anima 1.0 base as the pretrain to finetune into the JSON target. The upcoming JSON lora is being cached and trained with 40,000 of the full 83,000 valid images from the qwen set.
This first preview version is ready to use as a ComfyUI capable LORA, so you can just load up the epoch you want without anything special in comfyui and have at it. You can currently use plain English in conjunction with tagging to produce useful and meaningful prompt targets without the JSON.
AbstractPhil/anima-prelim-1k-r64
The comfyui nodes are present and work for testing use-case, but they are not ready for production use just yet.
-- Technical --
Primarily the target was the VLM json target followed by the AnimeTIMM vit processed through the VLM json processor as the followup. First 12 epochs VLM experienced images with json formatting, last 8 epochs were finetuning from epoch 12 onward to 20 using the AnimeTIMM captions turned into JSON instead.
The Anima model itself accepted the 1000 image and the json prompting works quite well. In the process I set up a couple comfyui nodes that can translate base prompts into the same language the model is learning. Those are present in the repo.
Full article available https://huggingface.co/blog/AbstractPhil/subject-bucketing.
There is additionally a civit model release as well.
https://civitai.com/models/2730503/anima-jsonenglish
AbstractPhil/anima-prelim-1k-r64
The JSON multi-prompt diffusion model prototype using Anima 1.0 base as the pretrain to finetune into the JSON target. The upcoming JSON lora is being cached and trained with 40,000 of the full 83,000 valid images from the qwen set.
This first preview version is ready to use as a ComfyUI capable LORA, so you can just load up the epoch you want without anything special in comfyui and have at it. You can currently use plain English in conjunction with tagging to produce useful and meaningful prompt targets without the JSON.
AbstractPhil/anima-prelim-1k-r64
The comfyui nodes are present and work for testing use-case, but they are not ready for production use just yet.
-- Technical --
Primarily the target was the VLM json target followed by the AnimeTIMM vit processed through the VLM json processor as the followup. First 12 epochs VLM experienced images with json formatting, last 8 epochs were finetuning from epoch 12 onward to 20 using the AnimeTIMM captions turned into JSON instead.
The Anima model itself accepted the 1000 image and the json prompting works quite well. In the process I set up a couple comfyui nodes that can translate base prompts into the same language the model is learning. Those are present in the repo.
updated 2
models 1 day ago
published a
model 1 day ago
updated a
collection 3 days ago
published a
model 3 days ago
updated a
dataset 3 days ago
liked a
model 5 days ago
updated a
model 7 days ago
published a
model 7 days ago
posted an update 8 days ago
Post
149
The article for aleph attention routing needs more work on vision, as the vision portion has not been fully validated, while the LM prototype has been semi-validated for small and medium-small scale. I will post my findings in the coming days with the consequences of training an LM and a VIT utilizing the prototype system.
The current structure for the Geometric Vocabulary does nearly reflect the intended shape as discussed in the earlier posts and articles, so that's coming along nicely - but there are stipulations and problems involved that I did not foresee.
My apologies for the incomplete article I just released on a whim. I jumped to the conclusion a bit early in anticipation before the formulas were fully converged. I also released an early post the other day speaking about the prototype AlephLM - which I removed as an invalid conclusion.
I'm doing my best to only release validated empirical information instead of speculative - however I do sometimes jump to conclusions without proper validation from time to time. Occasionally, I get a bit theory-overzealous and require tidying up through thorough experimentation which I'm currently approaching directly.
The current structure for the Geometric Vocabulary does nearly reflect the intended shape as discussed in the earlier posts and articles, so that's coming along nicely - but there are stipulations and problems involved that I did not foresee.
My apologies for the incomplete article I just released on a whim. I jumped to the conclusion a bit early in anticipation before the formulas were fully converged. I also released an early post the other day speaking about the prototype AlephLM - which I removed as an invalid conclusion.
I'm doing my best to only release validated empirical information instead of speculative - however I do sometimes jump to conclusions without proper validation from time to time. Occasionally, I get a bit theory-overzealous and require tidying up through thorough experimentation which I'm currently approaching directly.
updated a
model 10 days ago
published a
model 10 days ago
updated a
model 12 days ago
reacted to OzTianlu's post with 🧠 12 days ago
Post
6299
ResNet is Explicit Euler. GPT is Implicit Euler. What Else is Hiding in Plain Sight?
Read online: https://datawhalechina.github.io/learning-terrain/
I wrote an open-source monograph on learning dynamics — The Terrain of Learning. Bilingual (Chinese/English), 4 volumes, 12 chapters, 30+ print-grade figures. Completely free (CC BY-NC-SA 4.0).
The core argument: gradient descent is not optimization. It's terrain motion. The loss function is a landscape. The gradient is the direction of slope. The optimizer is how you choose each step. Once you see it this way, everything clicks:
ResNet = explicit Euler integration on a vector field. The residual branch is the vector field. Each layer takes one Euler step.
GPT autoregression = implicit-state Euler iteration. Stable where explicit Euler explodes. That's why transformers handle long-range dependencies.
DEQ = the Banach fixed-point theorem in production. The forward pass is root-finding. There are no layers to backprop through.
KL divergence = a Bregman divergence on the entropy landscape. Your belief space is curved, not flat.
Chain-of-thought reasoning = hidden states flowing along a reasoning field toward an attractor basin. Correct answers have wide basins. The number of reasoning steps is determined by the terrain, not by the problem.
Diffusion models = systems flowing downhill along a score vector field, from noise to structure, from high energy to low energy.
The book traces one idea across 337 years — from F=ma (Newton, 1687) to H=T+V (Hamilton, 1833) to loss landscape + gradient field (2020s). Hamilton replaced a catalog of forces with one geometric object. This book does the same for deep learning.
GitHub: https://github.com/datawhalechina/learning-terrain
Discussion: https://github.com/datawhalechina/learning-terrain/discussions/2
Convergence is not hope. Convergence is geometry. You see.
Read online: https://datawhalechina.github.io/learning-terrain/
I wrote an open-source monograph on learning dynamics — The Terrain of Learning. Bilingual (Chinese/English), 4 volumes, 12 chapters, 30+ print-grade figures. Completely free (CC BY-NC-SA 4.0).
The core argument: gradient descent is not optimization. It's terrain motion. The loss function is a landscape. The gradient is the direction of slope. The optimizer is how you choose each step. Once you see it this way, everything clicks:
ResNet = explicit Euler integration on a vector field. The residual branch is the vector field. Each layer takes one Euler step.
GPT autoregression = implicit-state Euler iteration. Stable where explicit Euler explodes. That's why transformers handle long-range dependencies.
DEQ = the Banach fixed-point theorem in production. The forward pass is root-finding. There are no layers to backprop through.
KL divergence = a Bregman divergence on the entropy landscape. Your belief space is curved, not flat.
Chain-of-thought reasoning = hidden states flowing along a reasoning field toward an attractor basin. Correct answers have wide basins. The number of reasoning steps is determined by the terrain, not by the problem.
Diffusion models = systems flowing downhill along a score vector field, from noise to structure, from high energy to low energy.
The book traces one idea across 337 years — from F=ma (Newton, 1687) to H=T+V (Hamilton, 1833) to loss landscape + gradient field (2020s). Hamilton replaced a catalog of forces with one geometric object. This book does the same for deep learning.
GitHub: https://github.com/datawhalechina/learning-terrain
Discussion: https://github.com/datawhalechina/learning-terrain/discussions/2
Convergence is not hope. Convergence is geometry. You see.
- 1 reply
geolip-aleph-void and the LM aleph routing is implicit recursive infinities confined into a microcosm of forced rebounding finite space forced through a gelu sift - more akin to an emulated quaternion. All because quaternion is computationally heavy and Cantor's fractals are additionally computationally precise (often >fp64 required), requiring an entirely deviant approach to rotary in order to computationally stabilize the system at BF16 so it won't take 2 weeks for a single epoch on a model 35m params.
Makes me feel a little overdressed for the occasion.
posted an update 13 days ago
Post
111
Claude Fable 5 was temp/perma? banned for security reasons.
Working with Fable I have to say the model is capable at handling highly complex geometric mathematics ACTUALLY to the point of me getting some work done without a headache. I hope Fable returns soon so I can finish cobbling without a headache and a week per prototype again.
During Fable's existence I managed to cobble together a multi-series aleph paradigm that can handle direct implicit and explicit learning for an LM with a trigram context window. This essentially provides expert directional utilization based on a stable codebook without requiring expert distillation into singular experts and duplicated.
Details soon. There are over 20 functional formula prototypes and around 8 potential heads that all lead to the same outcome, the math is rock solid - each with their own benefits and downsides based on the assigned text tasks.
Working with Fable I have to say the model is capable at handling highly complex geometric mathematics ACTUALLY to the point of me getting some work done without a headache. I hope Fable returns soon so I can finish cobbling without a headache and a week per prototype again.
During Fable's existence I managed to cobble together a multi-series aleph paradigm that can handle direct implicit and explicit learning for an LM with a trigram context window. This essentially provides expert directional utilization based on a stable codebook without requiring expert distillation into singular experts and duplicated.
Details soon. There are over 20 functional formula prototypes and around 8 potential heads that all lead to the same outcome, the math is rock solid - each with their own benefits and downsides based on the assigned text tasks.