Sensitive content
This media may contain sensitive content.
Загрузка видео...
Не удалось загрузить видео
Boobs work with a cunning cat! 😻❤️ Commission💋 🩷 Base model: DaB 🔉VA: No sound Also available on Patreon!🔞 #NSFW #rule34 #Cipher #CipherHSR #Hsr #HonkaiStarRail
287,358 просмотров • 7 месяцев назад •via X (Twitter)
Комментарии: 0
Нет доступных комментариев
Здесь появятся комментарии из оригинального поста
![I think I can finally report some success training a quite accurate IDM capable of recovering keystrokes from Minecraft gameplay, even in quite PvP-heavy situations. At this point the model does not only know what keys are pressed to the extent reasonably discernible, it also knows how fast it is moving in 3D space at all times, even when knockback is mixing with the self-move impulse. Now, recovering keystrokes from normal external capture footage is just about impossible. E.g. W/A/S/D does exactly nothing during partial tick frames and jumping mid-air is also equally useless, so asking the model to recover key down states is inherently unreasoanble. Mouse deltas are also completely arbitrary units, as game mouse sensitivity introduces an arbitrary scale factor into the equation. The only good option is to think carefully about your model-environment contract, and only record "logical actions", not raw keystrokes. So here's a few unfortunate lessons I had to learn in roughly this order. - Choose good units. (bad: mouse deltas, good: delta radians [yes, you will need game-internal state]) - Capture from inside the main game loop and read the game fbo to get consistent frame-action pairing. Doing post-mortem pairing is hopeless. - Carefully define when you think keystrokes actually have an effect. (jump only works on ground, when flying or in water etc.) More subtle: The key may already be down, but no tick has happened yet to actually use the value. Hence: ignore Seperate gamestate into "fast and slow-moving" components. E.g. movement is likely tick based, camera rotation is very likely updated every frame in essentially every game ever. - Think about your frame-action correspondance contract (How old is the frame in relation to the inputs you capture? Will double or tripple buffering affect you?) Think about the game loop timeline, where you are sampling, how old the data you are reading is, and where the ticks are happening around you. Language models used to simply not have a model-environment contract, but even now with the model "living" in a designated harness, the contract still boils down to formatting, and tool implementation intrinsics. While also important, it is still quite a bit more obvious because the violations are in some way shape or form reflected as text you can actually see. - ffmpeg dropping frames cummulatively screws the model the further you get into the sequence because your targets are now shifted. If you can't encode the video in real-time, too bad. - Sodium has a frames in flight system different from vanilla Minecraft, which will also offset your targets from your frames. (there goes that data...) - Models are succeptible to latency. If there is too big of a delay between action and on-screen reflection, your performance degrades. At this point I realize ~100hours of gameplay is essentially no longer usable as a dataset. You can train on this data, but all you'll get is a mushy mess. However, some good news: - Making the model predict physics gamestate scalars helps the model generalize. For instantaneous events like jump, it's unreasonable to ask the model emit a short burst of jump=true at exactly the right time, however if you also predict your current y-velocity, the model has supervision signal for the "latent" from which that onground jump becomes apparent. Recovering x/z motion is also somewhat easier than unmixing it into plausible keystrokes for inertia-heavy player controller logic. - Regressing physics gamestate scalars also seems to make your dataset "bigger". While pure keystroke classification will overfit quickly, predicting exact physics gamestate scalars forces the model to generalize more and you can tolerate far more epochs before validation loss starts to stall out. This is the only reason why it was bearable to dump 100h+ of dataset hours and replace it with ~3 hours of gameplay after the 4th revision of the file format (yeah...) and somehow still have better performance. Now, you might be asking, "isn't this brittle?" and the answer is yesn't. Frame-action correspondance matters for training, but not so much during inference. So as long as you are sampling in roughly the same interval as your training data, you aren't violating any hard contract per-se. Somewhere around the frames ticks are happening, and during training you capture various tick-capture offset relations per random chance, so nothing is too obviously wrong here. HOWEVER, you will get screwed by gui scale, shaders, resource packs, "shit that recording is 1920x1040 because somebody doesn't know fullscreen exists" and other unfortunate edge cases of reality. But I suppose this is the role of dataset size. If all those "contract violations" that a youtube video has compared to the training data are addressed, I think this is a way to turn Youtube into a labeled dataset. I could never shake the feeling that VPT is a sound idea in practice, while never having been properly executed, and I think one reason why it hasn't is because that label boostrapping part is just a pain in the butt to get right. Now, what the player is doing is of course not the only label you can extract from video, but it has to be one of the targets predicted during pretraining to "align" the pretraining objective. Some notes on the video here, the colored dots on the analog visualizer are the ground truth, while the gray dot is the model prediction. Green means correct prediction, red means incorrect prediction at that frame. Model P(key) reports how wrong the prediction is from green (0.0) to red (1.0). You will also notice that during periods of rapid slow down, left and right actions become close to irrecoverable, because there is just that little motion. And some jump actions are not predicted correctly because I got the detection condition for jump events wrong... (duh) LMB/RMB for other than sustained events (like item-consume and block break) also seem to be hopelessly irrecoverable for now. Swing was supposed to do the same thing as motion y did for jump, but its too well behaved as an increasing counter. Maybe partial-tick interpolated values work better (v5 file format then... ugh..)](https://image.24vids.com/tw-2042731804780826849/amplify_video_thumb/2042721691844120576/img/yugYQfuXXQmJpdVp.jpg)
![‼️🚨🔵PRESSER🔵🚨‼️ 🎙 Enzo Maresca’s Full Pre–Aston Villa Press Conference (Part 1 & 2) 🔹 Enzo Maresca on Delap and Estevão: “Both are available following their injuries.” ✅💪🏽 🔹 Enzo Maresca on if he would swap Morgan Rogers for Cole Palmer: “No. Because I love Cole ❤️ and Cole is my player.” 🔵 🔹 Enzo Maresca on Antoine Semenyo: “I’m focused on Aston Villa 🎯. Not on players who can arrive and leave. I’m in love with the squad we have.” 💙✨ 🔹 Enzo Maresca on Cole Palmer: “He is ready to play a full 90 minutes if needed.” ⏱️⚽ ⸻ 🔹 Enzo Maresca on any developments / the ‘worst 48 hours’ quote: “No, no, no. Before the Newcastle game, I already spoke about that. So nothing to add.” ❌🗣️ 🔹 Enzo Maresca on meetings with sporting directors / ownership: “Again, before the Newcastle game, in the press conference, I said that we had a meeting, that the relationship is good and we all focus on the Aston Villa game — the next game, the next game. So nothing to add.” 🤝🔵 🔹 Enzo Maresca on the players’ Christmas party: “Yeah, it was very good 🎄😊. Family time, players with their friends, with the kids. So it was a nice day, yeah.” 👨👩👧👦 🔹 Enzo Maresca on Chelsea’s training plan over Christmas: “After Newcastle they had two days off 😮💨. We trained yesterday, we train today, tomorrow off, and then Friday session and Saturday again.” 📆🏃🏽♂️ 🔹 Enzo Maresca on Pep giving players time off: “It’s Pep’s choice 🤝. He considers doing that. I think every manager probably thinks in a different way, but you have to respect different points of view. Is it about trusting the players to do the right thing on Christmas? No. I think it’s not about trust or not trust. I think also Pep trusts the players — just in a different way.” 🧠⚽ 🔹 Enzo Maresca on being in the stands vs Aston Villa: “Yes. In almost 100 games, I’ve been suspended three times 😅. One for celebration 🎉, and two times for yellow cards 🟨. So yeah, unfortunately I will be on the stand, but Willy [Caballero] is there, Danny [Walker] is there, the coaching staff is there. Does it affect the way I prepare, in terms of getting my messages to the players? No, no.” 🏟️📣 🔹 Enzo Maresca on facing Aston Villa: “It will be a very tough game for us ⚔️. We will have to be at our best because Aston Villa are playing very well. They have a clear way of playing and they are very organised both on and off the ball. They are full of confidence in this moment 🔥, but we are at home 🏠 and we will be trying to get the three points as we always do. We will work hard and continue to grow.” 📈💙](https://image.24vids.com/tw-2003805197680734348/amplify_video_thumb/2003804838019158016/img/bZv0h1VSVk2KXBks.jpg)
![Lamb and "John [Mack] were scheduled to deliver a paper...on reptilians." ~Ralph Blumenthal 👀🦎 John Mack, Reptilians and Death 🦎👀 "One woman worked for the [DEA] and was abducted while out on a snowmobile, during the day. Some thought the beings were robots. There were different races reported, including reptilians." ~~~ (Newer version of my 2022 thread about Blumenthal, Mack, and the paper Mack wrote on reptilians. Many will like this.) Ralph Blumenthal - Inquire Anomalous conference lecture from 12/03/2022. Notes... Blumenthal talks about a book he's written with his wife about UFOs that's geared towards children (I wonder how Semivan will feel about that? "They're f**king with us!" I have a contact who would REALLY not like that book). Blumenthal is spending his time going thru the story of John Mack and his work on abduction phenomenon. Mack tried LSD, ayahuasca and also holotropic breathing (taught by Stanislav Grof), where he seemed to have a possible past-life memory. Grof introduced Mack to UFOs and abductions. Mack turned down a chance to meet Budd Hopkins, but later, synchronistically, came across Hopkins again and they met in December 1990. Hopkins gave Mack some letters from experiencers and Mack decided to meet these folks and work with them. They were NOT mentally ill, deranged, or publicity seekers. It was so authentic that he couldn't believe they were making it up. The stories were surprisingly consistent and came from experiencers who were as young as three years old. Not all the events happened at night. One woman worked for the U.S. Drug Enforcement Agency and was abducted while out on a snowmobile, during the day. He also worked with women who claimed they were made pregnant, had the babies taken and then allowed to meet those children. Some thought the beings were robots. There were different races reported, including reptilians. The experiences were terrifying for many but also transformative and had a positive effect on the lives of some abductees. Mack's efforts to publish in peer-reviewed journals were rebuffed so he wrote books. He met with the Dalai Lama, who was interested in the abduction phenomenon. In 1994 or 1995, Mack hosted a conference at MIT on abductions. Lots of scientists attended and they discussed what they learned on abductions and shared notes. Mack thought these things must have happened on some level & Harvard (where he taught) was worried, so they tried to remove him. Mack hired a lawyer who didn't work out and he then hired a team that included Sheehan. One mistake Mack realized later was seeing abductions as singular when they were actually part of other mysteries/anomalies like cattle mutilations, ghosts, crop circles (which he investigated), ancient UFO sightings, fairies, NDEs, reincarnation, etc. Mack worked with a gifted scientist, Elizabeth Targ (Russell's daughter), who had special, intuitive powers. She was investigating brain tumors in her patients and wound up dying of the same brain tumor she was investigating. After she died, she somehow seemed able to communicate with her husband and family, from beyond. Mack was working on a book about her when he was killed. Blumenthal had access to all of the police files and came away convinced that Mack's death was a tragic accident. Barbara Lamb (abduction researcher/therapist) accompanied Mack on a visit to crop circles. Blumenthal mentions Lamb and reptilians (she allegedly had a close encounter in her home with a reptilian), and work Mack and Lamb were doing together on reptilians. But Mack died before that work was completed. Mack allegedly helped her from beyond the grave with that. I'll pull the video of that story (reptilians) and the last story and tweet it when the videos are available. "Death is not what we think it is." OK, a year later and I'm including the video clip (sorry for the sound issues) and here's the partial transcript... Blumenthal: "I tell the story of Barbara Lamb, who's a well known a psychotherapist, and experiencer, who had accompanied Mack on his visit to the crop circles. She was on the trip. And later, after Mack's death, Lamb was visiting her daughter in California, and with an allergy to her daughter's cat, she had an asthma attack that left her struggling to breathe. She thought she was suffocating, choking to death. And then she saw an image of Mack and he was saying, 'Don't worry, Barbara, you'll be okay.' And she felt a healing ball of light enter her chest and she suddenly could breathe. "Now, Lamb, who had intrigued me with a story of meeting a reptilian in her living room one day - and she had a very pleasant encounter, by the way - also told me that she and John were scheduled to deliver a paper at the Laughlin UFO conference on reptilians, and his death cut that short. And she didn't know what to do, whether she should still appear at the conference. "And John came to her in a dream, she told me, and said, by all means, she should still deliver his paper at the conference, and he told her where to find the paper. It was at home, on a certain bookshelf, the bookshelf nearest this other room, sticking out of certain papers. And that Barbara should call Karin Austin, who has Mack's assistant at the time, and describe just what John said, where the papers were. So Barbara Lamb did that, Karin located the papers, she delivered them to Barbara, Barbara delivered them at the conference and credited Mack with this report on reptilians." "It was not what we thought." ~Alleged message from a dead John Mack, via a medium. It? What is it? Death? UFOs? Clip is cued up so you can hear clean audio and the rest of the what Blumenthal said. Great lecture. #ufos #ufoX #aliens #uap #lifeafterdeath #afterlife #paranormal #healing #supernatural](https://image.24vids.com/tw-1740081390891470879/ext_tw_video_thumb/1740071055652446208/pu/img/qk1JlAWeQKdA6_yS.jpg)