Here's a bunch of TODOs to show that most of what I wanted to add I still haven't, regarding the work I did and the injustices done to me.
Overall, I'm still embarrassed by this repo and don't have the will to keep building something in a world where people can do what the University of Rochester institution, faculty members, and hospital staff did, who don't see the evil. None of you see the evil, and for that reason I don't want part in your world.
Maybe someday it'll be realized that all of you wasted something much more precious by torturing and tormenting me than I did by exiting. I leave my disproof of the Lorentz transformation to my uncles Slava and Sasha in my notebooks. My Github animal-tree has additional philosophy for those curious.
I wish you all revelations of your evils in unfriendly ways.
Quick ways to add detail
Systems note to niaz
Links to disease modeling and web app
Conversation with committee members + images
[ ] Formal theory process
[ ] Original XRD grant figures! - can add to 2nd link in indebted
[ ] And generalist multi-task email to committee members
I presented from this slide deck, generously available open-source and fully customizable:
[Here]() is the code I had already released open-source (in case my word and presentation alone weren’t good enough for them), prior to any of this including the committee meeting, independent from my [Builder]() code where I was privately progressing my research, including the committee member’s suggestion (a novel systems-level RL algorithm), that against all odds I successfully implemented, documented [in (e.g.) these commits](). (Extremely efficient distributed Replay prioritized batch sampling without replacement, followed by re-indexing upon online-RL data exhaustion).
My PhD committee terminated me for these ideas that I [had]() implemented and [have]() implemented. Institutions won’t support me, at least not the really almost-satirically dumb ones, and that’s a good thing, for the democratization and complete open-sourcing of everything I do with the public, though maybe not so good for privately-controlled ivy towers and centralized corporate powers. It means I need your help. Support [me]().
While Yann LeCun was proposing GEPA, I was building GEPA. While DeepMind was claiming a “generalist agent”, I was building one, including optimization. Well before transformers and ViTs gained popularity, I was innovating on them. My university never appreciated me. [I need you.]()
As a single individual who’s never made more than $15/hr, this was hard.
[After email]
These 3 slides summarize what I had already mostly finished implementing before I was terminated
I also presented this:
[Robotics diagram]
I needed a normal amount of time for a PhD program (7 years is common, even without two advisor changes and several medical malpractices) to finish what was literally the most important set of advancements in open-source ML research as of that date.
And I was terminated.
This essay makes an argument for panpsychism by reducing the alternative to the absurdity of a human-like conscious flipbook. I wrote the ideas simply so some of the most profound ones can be missed.
-> Add CS-based essay as well that I wrote.
Maybe qualia philosophy in something from nothing
Perhaps keeping order as is and appending to the end in Afterward, On the necessity of cyclicality
Make not just argument of cyclicality as a causal explanation of the universe, but alternatives including the infinitum limit, stochastic resampling of node subsets back to origin maybe, and more generally any cyclically connectable graph or whatever it is with the point that there also s no restriction against that graph expanding in number of vertical nodes (“bits”).
Maybe finally write the everything/anything/something argument and development stages. “Sam Lerman’s Cosmogenesis Theory”
Maybe qualia/observation philosophy in Consciousness.
“This is a conclusive disproof of most of the reasoning behind simulation-theory."
Be specific that we were doing X-ray diffraction modeling
This is what I showed to the committee, in May, 2023, most of it implemented (now available [here]()). They didn’t think there was anything novel in it. Here is the email conversation that followed afterward:
Ancillary
Don’t need to mention Kanan ‘Committee”
Diagram, modern slides, and emails
This was the rejected committee meeting that got me terminated, that lasted 20 minutes and involved brand new committee members who did not facilitate my research in any of my 6 years as a PhD student.
I did what he recommended, a lot of work by the way. Here is one of the commits that proves it:
I invented a novel and highly intuitive systems-level algorithm for reinforcement learning: (distributed/parallel) prioritized experience replay sampling without replacement, improving scores over non-partitioned sampling with replacement, and the first one to parallelize both with shared-RAM and memory-mapping systems allocation for SotA efficiency. That’s quite a lot top do on short-notice by one committee member’s email recommendation and then get terminated independently by my advisor without acknowledgement from the committee member!
And the termination, for not enough papers, was in the middle of a positive paper review of my second paper to get accepted and published in the 3 years I was under my advisor.
Emails
To: CS grad department
To: CS grad department students
In I built a whole GitHub for them, “including the whole open-access data pipeline and much of the data generation code and engineering and efficiency optimization”
Tell me in what way GATO is a research development if this is not, keeping in mind that this pre-dated GATO, and that GATO had all the resources of Google DeepMind, and I am one PhD student working across this and multiple other projects concurrently (by force), having never made more than $15/hr in my life, while battling actual induced medical tortures.
I was terminated a week before Sam Altman (my name is Sam Lerman), the head of OpenAI, was terminated by OpenAI. On the other hand, I suppose it’s good my AI & robot work is no longer in the hands of terminators.
And Chris Kanan couldn’t even remember the meeting, which means aside from Chenliang’s “not enough papers” reason, the decision ultimately came down to one of the other brand new committee members, Tom Howard of the _ department, who would definitely not accept this responsibility.
I could potentially add a README with summarizing diagrams. Part 1 - Sam Part 2 - Opposing Views …
Bio neuron can lead to ancillary with video image link play button and link and clarify with conversation below in text (not screen grabs) with “colleague” (if still saved) - coincidence by the way that made the jump but it wasn’t quite recorded (still witnessed) with Samgle
I can only guess retaliation for dropping the line. As far as they knew, I was in a hurry because the 30 minutes were up. Chenliang was also in a hurry. He told me he had to take his dog to the vet that day, or his wife would have to take him. They might’ve rescheduled, but as far as I knew, those 30 minutes were the cutoff point. He and his wife and probably vet and calendar can confirm.
I could maybe move the med malpractices to beginning of grads-alliance, with diagrams, of everything (waterboarding) and maybe photos of my arms from a while ago, and possibly videos (such as Mohsen’s) and decline
Maybe get stuff from emails about calling witnesses and the false accusations, such as of inconsistent research, numerical list of termination reasons with dates to sort when what was the termination reason
Termination reason 1: Not enough papers. I asked him if I was fired. He wouldn’t answer.
I can honestly say my eczema was regularly worse than this: (If I can’t find my own photos, find the worst one possible, centered below this)
Might be more readable if animated photos, unless of my own eczema
Maybe, at the end, after separator and in bold: Update: two months later I sent them this repo and they still haven’t replied. They either think I’m going to court or that the public won’t care. I’m not doing the first. I’m not sure how the public might react. I’ve only been ignored thus far.
Maybe: If that’s not malpractice, I don’t know what is.
After I lost my hair, we finally got a delivery of the drug. I wasn’t sure whether it’d be safer at that point to re-instigate the dependency/addiction or avoid the drug at all costs.
The I Ching said it didn’t matter. “The Well”, unchanging. It also said my hair would grow back of its own accord, which it did — on my head anyway — temporarily (when I did the all-steak diet amazingly). I still remember: “Fellowship with man” (e.g., “steak”, very manly) -> “Difficulty at the beginning” (“The name of the hexagram, Chun, really connotes a blade of grass pushing against an obstacle as it sprouts out of the earth”, https://www.google.com/search?client=safari&rls=en&q=i+ching+blades+of+grass&ie=UTF-8&oe=UTF-8). That answer about whether my hair’d grow back came true with the all-steak diet and my hair “sprouting” out of the earth, my head… temporarily.
When I did the all-steak diet again, a few eyelashes grew back. That can’t be a coincidence.
Add pictures of Bittle object recognizer
Bar separator below Michael
Maybe side-by-side images of the two deans above their paragraph, centered, tastefully medium-small
Add image of the Facebook post, clear and visible
-> …some of what -> occurred at the or “is occurring” or “is ongoing"
Terminator 2: Chris Kanan
Coincidence: The novel RL systems algorithm I innovated was in Experience Replay (memory, prioritized sampling)
Hey All, I’m running this by everybody to confirm if this is an accurate portrayal. Can I get a yay or nay?
Emphasize that the web app was good. It won the top 3 contest with a big cash prize.
Continue next parts starting from the next Part, e.g., Part 4, Part 5.
-
I couldn’t meet with them up until that point. Michael emailed about meeting with both him and Dean Nick Vamivakas. I replied: “”. Neither of them replied.
I also sent: “”.
In addition to Wendi and Nick not addressing me, Michael didn’t reply to any of my emails thereafter that mentioned anything substantive, not that he replied directly to anything I emailed before, including the request for evidence to back up their claims and [a simple yes/no question about whether their ultimate reason for termination was improvised]().
BioNeuron BioCell open-source model
[Here is a novel open-source model of GABA and Glutamate that I implemented some time ago, fully differentiable.]() to bottom of Benzodiazepine mechanism of action
I have consistently implemented the world’s most elegant [Architectures] link to them
We’re all grads
Add photo of burning alive to my… trauma
If I include references, can put in Ancillary, link as [1] or [2,3], and can number them with just hyperlinks, and under the hyperlinks can put a “Click here to expand citation” arrow button to expand citation
[Here]() is the accelerated and parallelized sampling without replacement for RL algorithm with memory-mapping and adaptive replay storage, and [here]() is one of the commits when I finished implementing the new systems RL algorithm,
They never replied. Here is a conclusive proof of my arguments:
Call the disproof Disproof, put disproof last. Replies first to first emails, “similar experience” and Curtis and jaeger, any other disproving emails, and anti-evidence like group witnesses
Can link the commit where I did the multi-task. Might be in unifiedml-legacy
Ancillary Emails can be referenced like references [[1]]() with for example title: “# 1.” and then email, then “# 2.” and so on.One file for Chris Kanan’s second wave of emails, I think
Make all images links to full screen
Maybe make dark bg versions as well
PhD students are employees
Employees are not slaves
Slaves are involuntary workers whose work has not yet been paid for - it can be paid for later via reparations
To pay for work requires a living wage
University of Rochester’s PhD student stipend is below New York state’s official living wage
Voluntary vs. involuntary is evidenced by what people would choose to do without guarantee of pay
Add previous draft versions to Unused Slides Or maybe call it Version Control
Add the task= to the modern unifiedml slides xrd multi task example commands
Maybe can add Michael Scott and group synchronicities to Faith
Including grade bump and, maybe, vaguely, Kanan (multiple times and around the theme of memory crucially)
And Jordan Peterson and “Fellowship with man” / Mikhaila Peterson maybe and diet (teeth symbolism? especially with “low-confidence” mentioned around disobeying outpatient psychiatrist)
Sam Altman
They are currently ghosting me (as of ).
The combination of some of those abstinences with fasting was really good for my eczema
Here's a bunch of TODOs to show that most of what I wanted to add I still haven't, regarding the work I did and the injustices done to me.
Overall, I'm still embarrassed by this repo and don't have the will to keep building something in a world where people can do what the University of Rochester institution, faculty members, and hospital staff did, who don't see the evil. None of you see the evil, and for that reason I don't want part in your world.
Maybe someday it'll be realized that all of you wasted something much more precious by torturing and tormenting me than I did by exiting. I leave my disproof of the Lorentz transformation to my uncles Slava and Sasha in my notebooks. My Github animal-tree has additional philosophy for those curious.
I wish you all revelations of your evils in unfriendly ways.
Quick ways to add detail
Systems note to niaz
Links to disease modeling and web app
Conversation with committee members + images
[ ] Formal theory process
[ ] Original XRD grant figures! - can add to 2nd link in indebted
I presented from this slide deck, generously available open-source and fully customizable:
[Here]() is the code I had already released open-source (in case my word and presentation alone weren’t good enough for them), prior to any of this including the committee meeting, independent from my [Builder]() code where I was privately progressing my research, including the committee member’s suggestion (a novel systems-level RL algorithm), that against all odds I successfully implemented, documented [in (e.g.) these commits](). (Extremely efficient distributed Replay prioritized batch sampling without replacement, followed by re-indexing upon online-RL data exhaustion).
Single-task: Multi-task: Robotics: Fully documented open-sourced legacy state:
My PhD committee terminated me for these ideas that I [had]() implemented and [have]() implemented. Institutions won’t support me, at least not the really almost-satirically dumb ones, and that’s a good thing, for the democratization and complete open-sourcing of everything I do with the public, though maybe not so good for privately-controlled ivy towers and centralized corporate powers. It means I need your help. Support [me]().
While Yann LeCun was proposing GEPA, I was building GEPA. While DeepMind was claiming a “generalist agent”, I was building one, including optimization. Well before transformers and ViTs gained popularity, I was innovating on them. My university never appreciated me. [I need you.]()
As a single individual who’s never made more than $15/hr, this was hard.
[After email] These 3 slides summarize what I had already mostly finished implementing before I was terminated
I also presented this: [Robotics diagram]
I needed a normal amount of time for a PhD program (7 years is common, even without two advisor changes and several medical malpractices) to finish what was literally the most important set of advancements in open-source ML research as of that date.
And I was terminated.
This essay makes an argument for panpsychism by reducing the alternative to the absurdity of a human-like conscious flipbook. I wrote the ideas simply so some of the most profound ones can be missed.
-> Add CS-based essay as well that I wrote.
Maybe qualia philosophy in something from nothing
“This is a conclusive disproof of most of the reasoning behind simulation-theory."
Be specific that we were doing X-ray diffraction modeling
This is what I showed to the committee, in May, 2023, most of it implemented (now available [here]()). They didn’t think there was anything novel in it. Here is the email conversation that followed afterward: Ancillary Don’t need to mention Kanan ‘Committee” Diagram, modern slides, and emails This was the rejected committee meeting that got me terminated, that lasted 20 minutes and involved brand new committee members who did not facilitate my research in any of my 6 years as a PhD student.
I did what he recommended, a lot of work by the way. Here is one of the commits that proves it: I invented a novel and highly intuitive systems-level algorithm for reinforcement learning: (distributed/parallel) prioritized experience replay sampling without replacement, improving scores over non-partitioned sampling with replacement, and the first one to parallelize both with shared-RAM and memory-mapping systems allocation for SotA efficiency. That’s quite a lot top do on short-notice by one committee member’s email recommendation and then get terminated independently by my advisor without acknowledgement from the committee member! And the termination, for not enough papers, was in the middle of a positive paper review of my second paper to get accepted and published in the 3 years I was under my advisor.
Emails To: CS grad department To: CS grad department students In I built a whole GitHub for them, “including the whole open-access data pipeline and much of the data generation code and engineering and efficiency optimization”
Tell me in what way GATO is a research development if this is not, keeping in mind that this pre-dated GATO, and that GATO had all the resources of Google DeepMind, and I am one PhD student working across this and multiple other projects concurrently (by force), having never made more than $15/hr in my life, while battling actual induced medical tortures.
Both of these are the urgsu website if needed: https://github.com/URGSU/TempURGSU and https://github.com/URGSU/Website
I was terminated a week before Sam Altman (my name is Sam Lerman), the head of OpenAI, was terminated by OpenAI. On the other hand, I suppose it’s good my AI & robot work is no longer in the hands of terminators.
And Chris Kanan couldn’t even remember the meeting, which means aside from Chenliang’s “not enough papers” reason, the decision ultimately came down to one of the other brand new committee members, Tom Howard of the _ department, who would definitely not accept this responsibility.
I could potentially add a README with summarizing diagrams. Part 1 - Sam Part 2 - Opposing Views …
Bio neuron can lead to ancillary with video image link play button and link and clarify with conversation below in text (not screen grabs) with “colleague” (if still saved) - coincidence by the way that made the jump but it wasn’t quite recorded (still witnessed) with Samgle
I can only guess retaliation for dropping the line. As far as they knew, I was in a hurry because the 30 minutes were up. Chenliang was also in a hurry. He told me he had to take his dog to the vet that day, or his wife would have to take him. They might’ve rescheduled, but as far as I knew, those 30 minutes were the cutoff point. He and his wife and probably vet and calendar can confirm.
I could maybe move the med malpractices to beginning of grads-alliance, with diagrams, of everything (waterboarding) and maybe photos of my arms from a while ago, and possibly videos (such as Mohsen’s) and decline
Maybe get stuff from emails about calling witnesses and the false accusations, such as of inconsistent research, numerical list of termination reasons with dates to sort when what was the termination reason
Termination reason 1: Not enough papers. I asked him if I was fired. He wouldn’t answer.
I can honestly say my eczema was regularly worse than this: (If I can’t find my own photos, find the worst one possible, centered below this)
Maybe, at the end, after separator and in bold: Update: two months later I sent them this repo and they still haven’t replied. They either think I’m going to court or that the public won’t care. I’m not doing the first. I’m not sure how the public might react. I’ve only been ignored thus far.
Maybe: If that’s not malpractice, I don’t know what is.
After I lost my hair, we finally got a delivery of the drug. I wasn’t sure whether it’d be safer at that point to re-instigate the dependency/addiction or avoid the drug at all costs.
The I Ching said it didn’t matter. “The Well”, unchanging. It also said my hair would grow back of its own accord, which it did — on my head anyway — temporarily (when I did the all-steak diet amazingly). I still remember: “Fellowship with man” (e.g., “steak”, very manly) -> “Difficulty at the beginning” (“The name of the hexagram, Chun, really connotes a blade of grass pushing against an obstacle as it sprouts out of the earth”, https://www.google.com/search?client=safari&rls=en&q=i+ching+blades+of+grass&ie=UTF-8&oe=UTF-8). That answer about whether my hair’d grow back came true with the all-steak diet and my hair “sprouting” out of the earth, my head… temporarily.
When I did the all-steak diet again, a few eyelashes grew back. That can’t be a coincidence.
Add pictures of Bittle object recognizer
Bar separator below Michael
Maybe side-by-side images of the two deans above their paragraph, centered, tastefully medium-small
Add image of the Facebook post, clear and visible
-> …some of what -> occurred at the or “is occurring” or “is ongoing"
Terminator 2: Chris Kanan Coincidence: The novel RL systems algorithm I innovated was in Experience Replay (memory, prioritized sampling) Hey All, I’m running this by everybody to confirm if this is an accurate portrayal. Can I get a yay or nay?
Emphasize that the web app was good. It won the top 3 contest with a big cash prize.
Continue next parts starting from the next Part, e.g., Part 4, Part 5.
-
I couldn’t meet with them up until that point. Michael emailed about meeting with both him and Dean Nick Vamivakas. I replied: “”. Neither of them replied.
I also sent: “”.
In addition to Wendi and Nick not addressing me, Michael didn’t reply to any of my emails thereafter that mentioned anything substantive, not that he replied directly to anything I emailed before, including the request for evidence to back up their claims and [a simple yes/no question about whether their ultimate reason for termination was improvised]().
BioNeuron BioCell open-source model
[Here is a novel open-source model of GABA and Glutamate that I implemented some time ago, fully differentiable.]() to bottom of Benzodiazepine mechanism of action
I have consistently implemented the world’s most elegant [Architectures] link to them
We’re all grads
Add photo of burning alive to my… trauma If I include references, can put in Ancillary, link as [1] or [2,3], and can number them with just hyperlinks, and under the hyperlinks can put a “Click here to expand citation” arrow button to expand citation
[Here]() is the accelerated and parallelized sampling without replacement for RL algorithm with memory-mapping and adaptive replay storage, and [here]() is one of the commits when I finished implementing the new systems RL algorithm,
They never replied. Here is a conclusive proof of my arguments:
Call the disproof Disproof, put disproof last. Replies first to first emails, “similar experience” and Curtis and jaeger, any other disproving emails, and anti-evidence like group witnesses
Can link the commit where I did the multi-task. Might be in unifiedml-legacy
Ancillary Emails can be referenced like references [[1]]() with for example title: “# 1.” and then email, then “# 2.” and so on.One file for Chris Kanan’s second wave of emails, I think
Make all images links to full screen Maybe make dark bg versions as well
PhD students are employees Employees are not slaves Slaves are involuntary workers whose work has not yet been paid for - it can be paid for later via reparations To pay for work requires a living wage University of Rochester’s PhD student stipend is below New York state’s official living wage Voluntary vs. involuntary is evidenced by what people would choose to do without guarantee of pay
Add previous draft versions to Unused Slides Or maybe call it Version Control
Add the task= to the modern unifiedml slides xrd multi task example commands
Maybe can add Michael Scott and group synchronicities to Faith Including grade bump and, maybe, vaguely, Kanan (multiple times and around the theme of memory crucially) And Jordan Peterson and “Fellowship with man” / Mikhaila Peterson maybe and diet (teeth symbolism? especially with “low-confidence” mentioned around disobeying outpatient psychiatrist) Sam Altman
They are currently ghosting me (as of).
The combination of some of those abstinences with fasting was really good for my eczema
Bittle commits I think https://github.com/slerman12/Builder/commits/main/?after=64c92d7bf5c09642e161548c36625739884b01f7+69