Video yükleniyor...

Video Yüklenemedi

Bu video yüklenirken bir sorun oluştu. Bu geçici bir ağ sorunundan kaynaklanıyor olabilir veya video kullanılamıyor olabilir.

Ana Sayfaya Dön

Gray CTR bones key animated & green DEF bones apply corrections. 1. Simple flat hidden mesh, using hook modifiers to follow the CTR bones. 2. Hidden mesh uses "outside" shrinkwrap to conform to sphere (or any object) 3. DEF bones are damp tracked to vertex groups on hidden mesh

snek3d

7,366 subscribers

27,434 görüntüleme • 6 ay önce •via X (Twitter)

Anya Rossi• Live Now

Private livecam show

0 Yorum

Yorum bulunmuyor

Orijinal gönderinin yorumları burada görünecek

Benzer Videolar

A Beverage that your gut will thank you for. BONE BROTH is simple to prepare. 1. Roast the Bones [ I have used Goat Trotters & Rear Leg Marrow Bones] 2. Add Apple Cider Vinegar or Vinegar. 3. Add Spices & Aromatics of your choice. 4. Add Water. 5. Pressure Cook or Slow Cook.

A Beverage that your gut will thank you for. BONE BROTH is simple to prepare. 1. Roast the Bones [ I have used Goat Trotters & Rear Leg Marrow Bones] 2. Add Apple Cider Vinegar or Vinegar. 3. Add Spices & Aromatics of your choice. 4. Add Water. 5. Pressure Cook or Slow Cook.

Anand Nagu

16,591 görüntüleme • 10 ay önce

EASIEST BONE BROTH RECIPE on the Internet. 1. Start with adding some Apple Cider Vinegar to Goat Trotters(or whatever Bones you have/prefer) 2. Add your choice of spices, herbs & aromatics. 3. Add Three times the water as the Bones used. 4. Add Salt. 5. It’s Slow cooked under a lid Preferably for atleast 8 hours for the best possible mineral & collagen extraction.

EASIEST BONE BROTH RECIPE on the Internet. 1. Start with adding some Apple Cider Vinegar to Goat Trotters(or whatever Bones you have/prefer) 2. Add your choice of spices, herbs & aromatics. 3. Add Three times the water as the Bones used. 4. Add Salt. 5. It’s Slow cooked under a lid Preferably for atleast 8 hours for the best possible mineral & collagen extraction.

Anand Nagu

21,817 görüntüleme • 8 ay önce

[LSTM] by Hand ✍️ LSTMs have been the most effective architecture to process long sequences of data, until our world was taken over by the Transformers. LSTMs belong to the broader family of recurrent neural network (RNNs) that process data sequentially in a recurrent manner. Transformers, on the other hand, abandon recurrence and use self-attention instead to process data concurrently in parallel. Recently, there is renewed interest in recurrence as people realized self-attention doesn’t scale to extremely long sequences, like hundreds of thousands of tokens. Mamba is a good example to bring back recurrence. All of a sudden, it is cool to study LSTMs. How do LSTMs work? [1] Given ↳ 🟨 Input sequence X1, X2, X3 (d = 3) ↳ 🟩 Hidden state h (d = 2) ↳ 🟦 Memory C (d = 2) ↳ Weight matrices Wf, Wc, Wi, Wo Process t = 1 [2] Initialize ↳ Randomly set the previous hidden state h0 to [1, 1] and memory cells C0 to [0.3, -0.5] [3] Linear Transform ↳ Multiply the four weight matrices with the concatenation of current input (X1) and the previous hidden state (h0). ↳ The results are feature values, each is a linear combination of the current input and hidden state. [4] Non-linear Transform ↳ Apply sigmoid σ to obtain gate values (between 0 and 1). • Forget gate (f1): [-4, -6] → [0, 0] • Input gate (i1): [6, 4] → [1, 1] • Output gate (o1): [4, -5] → [1, 0] ↳ Apply tanh to obtain candidate memory values (between -1 and 1) • Candidate memory (C’1): [1, -6] → [0.8, -1] [5] Update Memory ↳ Forget (C0 .* f1): Element-wise multiply the current memory with forget gate values. ↳ Input (C’1 .* o1): Element-wise multiply the “candidate” memory with input gate values. ↳ Update the memory to C1 by adding the two terms above: C0 .* f1 + C’1 .* o1 = C1 [6] Candiate Output ↳ Apply tanh to the new memory C1 to obtain candidate output o’1. [0.8, -1] → [0.7, -0.8] [7] Update Hidden State ↳ Output (o’1 .* o1 → h1): Element-wise multiply the candidate output with the output gate. ↳ The result is updated hidden state h1 ↳ Also, it is the first output. Process t = 2 [8] Initialize ↳ Copy previous hidden state h1 and memory C1 [9] Linear Transform ↳ Repeat [3] [10] Update Memory (C2) ↳ Repeat [4] and [5] [11] Update Hidden State (h2) ↳ Repeat [6] and [7] Process t = 3 [12] Initialize ↳ Copy previous hidden state h2 and memory C2 [13] Linear Transform ↳ Repeat [3] [14] Update Memory (C3) ↳ Repeat [4] and [5] [15] Update Hidden State (h3) ↳ Repeat [6] and [7]

[LSTM] by Hand ✍️ LSTMs have been the most effective architecture to process long sequences of data, until our world was taken over by the Transformers. LSTMs belong to the broader family of recurrent neural network (RNNs) that process data sequentially in a recurrent manner. Transformers, on the other hand, abandon recurrence and use self-attention instead to process data concurrently in parallel. Recently, there is renewed interest in recurrence as people realized self-attention doesn’t scale to extremely long sequences, like hundreds of thousands of tokens. Mamba is a good example to bring back recurrence. All of a sudden, it is cool to study LSTMs. How do LSTMs work? [1] Given ↳ 🟨 Input sequence X1, X2, X3 (d = 3) ↳ 🟩 Hidden state h (d = 2) ↳ 🟦 Memory C (d = 2) ↳ Weight matrices Wf, Wc, Wi, Wo Process t = 1 [2] Initialize ↳ Randomly set the previous hidden state h0 to [1, 1] and memory cells C0 to [0.3, -0.5] [3] Linear Transform ↳ Multiply the four weight matrices with the concatenation of current input (X1) and the previous hidden state (h0). ↳ The results are feature values, each is a linear combination of the current input and hidden state. [4] Non-linear Transform ↳ Apply sigmoid σ to obtain gate values (between 0 and 1). • Forget gate (f1): [-4, -6] → [0, 0] • Input gate (i1): [6, 4] → [1, 1] • Output gate (o1): [4, -5] → [1, 0] ↳ Apply tanh to obtain candidate memory values (between -1 and 1) • Candidate memory (C’1): [1, -6] → [0.8, -1] [5] Update Memory ↳ Forget (C0 .* f1): Element-wise multiply the current memory with forget gate values. ↳ Input (C’1 .* o1): Element-wise multiply the “candidate” memory with input gate values. ↳ Update the memory to C1 by adding the two terms above: C0 .* f1 + C’1 .* o1 = C1 [6] Candiate Output ↳ Apply tanh to the new memory C1 to obtain candidate output o’1. [0.8, -1] → [0.7, -0.8] [7] Update Hidden State ↳ Output (o’1 .* o1 → h1): Element-wise multiply the candidate output with the output gate. ↳ The result is updated hidden state h1 ↳ Also, it is the first output. Process t = 2 [8] Initialize ↳ Copy previous hidden state h1 and memory C1 [9] Linear Transform ↳ Repeat [3] [10] Update Memory (C2) ↳ Repeat [4] and [5] [11] Update Hidden State (h2) ↳ Repeat [6] and [7] Process t = 3 [12] Initialize ↳ Copy previous hidden state h2 and memory C2 [13] Linear Transform ↳ Repeat [3] [14] Update Memory (C3) ↳ Repeat [4] and [5] [15] Update Hidden State (h3) ↳ Repeat [6] and [7]

Tom Yeh

72,966 görüntüleme • 2 yıl önce

質問来たので貫通対策についてご紹介します！ ①upperlegの子にちかくのスカートのボーンを入れる。 →スカート前方に2本置いているのは子にした時に不自然な形にならない位置に調整するため。またボーン数が少ない事で、スカート前方ボーンからサイドのボーンまでの変形が緩やかに起こるため形が綺麗になる。 ②角度制限をつけて揺らす。 →upperlegの子のスカートボーン、第1ボーンは揺れの角度を規定(↓水色のエリア)する事でupperlegとの干渉が物理的に起こりえなくなる。 ③第2ボーンは重力設定とコライダーで自然な動きにします。第2ボーンに重力設定をすることで座った時の形、足を折り曲げた時に膝に引っかかっている表現をしています。この表現のためにスカートボーンの関節を膝関節よりやや下にすると綺麗に動作します。そこそこ貫通に強いです✨ I received a question, so I'd like to share how I handle clipping prevention! ① Add skirt bones near the upper leg as child bones. → The two bones placed at the front of the skirt are positioned in a way that avoids unnatural shapes when they become children of the upper leg. Also, by keeping the number of bones low, the deformation from the front skirt bones to the side bones becomes more gradual, resulting in a cleaner shape. ② Apply angle limits and simulate sway. → For the first skirt bone (child of the leg), limiting the sway angle (see the light blue area below) ensures it can’t physically collide with the upper leg. ③ Use gravity settings and colliders on the second bone for natural movement. → By setting gravity on the second bone, you can express the skirt catching on the knees when sitting or bending the legs. To achieve this effect, placing the joint of the skirt bone slightly below the knee joint allows for smoother movement. It's fairly resistant to clipping ✨ #blender #VRChat #3DCG

質問来たので貫通対策についてご紹介します！ ①upperlegの子にちかくのスカートのボーンを入れる。 →スカート前方に2本置いているのは子にした時に不自然な形にならない位置に調整するため。またボーン数が少ない事で、スカート前方ボーンからサイドのボーンまでの変形が緩やかに起こるため形が綺麗になる。 ②角度制限をつけて揺らす。 →upperlegの子のスカートボーン、第1ボーンは揺れの角度を規定(↓水色のエリア)する事でupperlegとの干渉が物理的に起こりえなくなる。 ③第2ボーンは重力設定とコライダーで自然な動きにします。第2ボーンに重力設定をすることで座った時の形、足を折り曲げた時に膝に引っかかっている表現をしています。この表現のためにスカートボーンの関節を膝関節よりやや下にすると綺麗に動作します。そこそこ貫通に強いです✨ I received a question, so I'd like to share how I handle clipping prevention! ① Add skirt bones near the upper leg as child bones. → The two bones placed at the front of the skirt are positioned in a way that avoids unnatural shapes when they become children of the upper leg. Also, by keeping the number of bones low, the deformation from the front skirt bones to the side bones becomes more gradual, resulting in a cleaner shape. ② Apply angle limits and simulate sway. → For the first skirt bone (child of the leg), limiting the sway angle (see the light blue area below) ensures it can’t physically collide with the upper leg. ③ Use gravity settings and colliders on the second bone for natural movement. → By setting gravity on the second bone, you can express the skirt catching on the knees when sitting or bending the legs. To achieve this effect, placing the joint of the skirt bone slightly below the knee joint allows for smoother movement. It's fairly resistant to clipping ✨ #blender #VRChat #3DCG

しろいの〔VRCアバター制作〕

55,393 görüntüleme • 1 yıl önce

what the fuck.. Cargill is using AI cameras in *slaughterhouses* to extract millions of dollars of extra meat off the bones of animals.. here’s the TLDR: >workers get real-time smiley face ratings per cut. green, yellow, red. > 1% improvement = 200 million more pounds of beef per year. >proprietary AI camera system called “CarVe” >scans every cut in real time on the fabrication floor >flags meat left on the bone that workers missed >rates each cut green, yellow, or red so workers adjust instantly >piloted in Friona, TX. now rolling out across all plants this is not a tech company. this is a slaughterhouse in Texas. and AI is already grading every cut in real time. this is insanity.

what the fuck.. Cargill is using AI cameras in slaughterhouses to extract millions of dollars of extra meat off the bones of animals.. here’s the TLDR: >workers get real-time smiley face ratings per cut. green, yellow, red. > 1% improvement = 200 million more pounds of beef per year. >proprietary AI camera system called “CarVe” >scans every cut in real time on the fabrication floor >flags meat left on the bone that workers missed >rates each cut green, yellow, or red so workers adjust instantly >piloted in Friona, TX. now rolling out across all plants this is not a tech company. this is a slaughterhouse in Texas. and AI is already grading every cut in real time. this is insanity.

Damian Player

843,444 görüntüleme • 5 ay önce

While Prince Harry and Meghan Markle (who never provide any value to the world) enjoy their mansions, courtside tickets, and celebrity meet n’ greets, there are real problems that real Americans are facing. Companies like Hims and others are still trying to sell people on using potentially dangerous copycat GLP-1 weight loss drugs. Hidden untested Chinese ingredients should not be flooding American markets. Not that royal 1 and royal 2 over there would care, but making America healthy again includes protection from foreign actors. The FDA cracked down on Hims, but it shouldn’t stop there. It’s time for them to go after Ro, Willow, and every other company mass-marketing compounded GLP-1s without FDA approval.

While Prince Harry and Meghan Markle (who never provide any value to the world) enjoy their mansions, courtside tickets, and celebrity meet n’ greets, there are real problems that real Americans are facing. Companies like Hims and others are still trying to sell people on using potentially dangerous copycat GLP-1 weight loss drugs. Hidden untested Chinese ingredients should not be flooding American markets. Not that royal 1 and royal 2 over there would care, but making America healthy again includes protection from foreign actors. The FDA cracked down on Hims, but it shouldn’t stop there. It’s time for them to go after Ro, Willow, and every other company mass-marketing compounded GLP-1s without FDA approval.

USA NEWS 🇺🇸

235,285 görüntüleme • 6 ay önce

Facial reconstruction of a 8,000-year-old EHG from Karelia The individual (Oleny Ostrov grave 75), who carried Y-DNA J1, was found in the famous Neolithic cemetary of Oleny Ostrov. The Oleny Ostrov burial ground is a vast Neolithic necropolis on South Oleny Island in Lake Onega, Karelia. Excavated between 1936 and 1938 by Professor V. I. Ravdonikas, it is one of the most important prehistoric cemeteries in northern Europe. Its significance lies in its large number of burials, exceptionally rich archaeological and anthropological remains, and a distinctive funerary tradition unlike any other known Neolithic cemetery. The site is also notable for its remarkable bone artifacts and unique stone tool assemblage. The individual was described by the anthropologist V.P. Yakimov (1960) as follows: The skeleton was found lying on its back in an extended position. It was oriented with the head toward the east, slightly deviating toward the north. No grave goods were present. Determining the sex is rather difficult because the diagnostic characteristics are only weakly expressed. The individual was most likely male, although some uncertainty remains. The individual died at a mature adult age. The skull is medium-sized, elongated, and heavy, with very thick cranial bones but only moderate brow ridges and cranial relief. The forehead is moderately sloping, the cranial vault long with flattened parietals, and the occiput bears a low occipital torus. The face is relatively high and of medium breadth. It exhibits considerable flatness in both the orbital and zygomatic regions. The canine fossae are moderately expressed (grade 2). The cheekbones are low. In frontal view, the contour of the zygomatic bones is intermediate between the classic "Europoid" form, with a pronounced hook at the zygomaxillary suture, and the "Mongolid" form, where the transition between the maxillary and zygomatic bones is smooth and lacks such a hook. The nasal aperture is narrow, prominent but flattened-root nasal bones, and an orthognathous profile. The mandible is medium-sized but robust, with a thick body, broad rami, and a moderately projecting chin. The teeth are relatively small with light to moderate wear. The limb bones are generally slender with weak to moderate muscle markings, although the humeri are relatively robust. The clavicles are long, slender, and strongly curved. The individual (UOO059) carried Y-DNA J1 and mtDNA R1b. His maternal lineage has been identified in the ANE individual Afontova Gora 3, while his Y-DNA lineage has been found in several other EHG individuals, including two other at Oleny Ostrov and one at Popovo nearby. Although its ultimate origin remains uncertain. Its relationship to the J1 lineage from Satsurblia may suggest that it spread from the Caucasus to Karelia along the Volga. Autosomally, he was a typical Karelian EHG. Karelian EHGs differed slightly from the main Sidelkino EHG cluster by harboring additional ANE/less WHG admixture.

Ancestral Whispers

37,985 görüntüleme • 2 ay önce

a🧵on how to do the movable crater tech, You will learn this step by step, It should only take a couple of minutes to do it so lock in, Simple Overview: 1) Draw a black texture that represents a crack (use lasso or draw then fill tool) 2) Use a png to svg converter to make it into svg (you MUST make the crack black!) 3) Make the mesh of the crater 4) Export it as fbx and Import it in Roblox Studio 5) Create a viewport put the thing inside play with some lighting,material of crater n ur done 6) Make another viewport and make it so that it duplicates the character of the player every frame and puts in the viewport so that you can see it to be above the crater viewport Now for the detailed tutorial:

a🧵on how to do the movable crater tech, You will learn this step by step, It should only take a couple of minutes to do it so lock in, Simple Overview: 1) Draw a black texture that represents a crack (use lasso or draw then fill tool) 2) Use a png to svg converter to make it into svg (you MUST make the crack black!) 3) Make the mesh of the crater 4) Export it as fbx and Import it in Roblox Studio 5) Create a viewport put the thing inside play with some lighting,material of crater n ur done 6) Make another viewport and make it so that it duplicates the character of the player every frame and puts in the viewport so that you can see it to be above the crater viewport Now for the detailed tutorial:

QwinkleTee

40,103 görüntüleme • 1 yıl önce

[WIP] Hypno & Outline shaders converted over to Unity.

[WIP] Hypno & Outline shaders converted over to Unity.

snek3d

24,729 görüntüleme • 2 ay önce

One issue I truly hope will finally be correctly addressed post revolution, is the stray dog situation in Iran. Iran has absolutely beautiful dogs. They are gentle, affectionate, and remarkably calm. I’m the type of person who will run up to a dog and start petting it without hesitation, often forgetting that a dog could react aggressively, especially in Iran, where many of these poor animals have endured harsh lives. Yet I have never had a single negative experience with stray dogs in Iran. I have carried them, washed them, and cared for them, and not once did they show any sign of aggression. There are compassionate individuals trying to help where they can. Some people have built small informal networks with local butchers to collect leftover bones and scraps so they can at least feed the dogs. These efforts come from genuine care and personal sacrifice. But this is not a structural solution. It is only a bandage on a deep wound. The current way the government handles the stray population is cruel and inhumane. Dogs are captured, their mouths are taped shut, and they are left to slowly starve. At the same time, ordinary people face heavy restrictions on keeping dogs. Small dogs can sometimes be hidden inside apartments, but larger ones are nearly impossible to keep. If you walk your dog outside, you risk it being confiscated. Even inside your own home, a hostile neighbor can report you simply for owning a dog, and authorities can act on that complaint. This creates an environment where ownership is punished, while the real problem continues unchecked on the streets. What Iran needs is a humane, practical, and modern solution: 1. A nationwide Catch–Neuter–Vaccinate–Release (CNVR) program 2. Properly managed municipal shelters with medical standards 3. Public education and cultural change We have many urgente matters to attend to after this vile and evil regime falls.

One issue I truly hope will finally be correctly addressed post revolution, is the stray dog situation in Iran. Iran has absolutely beautiful dogs. They are gentle, affectionate, and remarkably calm. I’m the type of person who will run up to a dog and start petting it without hesitation, often forgetting that a dog could react aggressively, especially in Iran, where many of these poor animals have endured harsh lives. Yet I have never had a single negative experience with stray dogs in Iran. I have carried them, washed them, and cared for them, and not once did they show any sign of aggression. There are compassionate individuals trying to help where they can. Some people have built small informal networks with local butchers to collect leftover bones and scraps so they can at least feed the dogs. These efforts come from genuine care and personal sacrifice. But this is not a structural solution. It is only a bandage on a deep wound. The current way the government handles the stray population is cruel and inhumane. Dogs are captured, their mouths are taped shut, and they are left to slowly starve. At the same time, ordinary people face heavy restrictions on keeping dogs. Small dogs can sometimes be hidden inside apartments, but larger ones are nearly impossible to keep. If you walk your dog outside, you risk it being confiscated. Even inside your own home, a hostile neighbor can report you simply for owning a dog, and authorities can act on that complaint. This creates an environment where ownership is punished, while the real problem continues unchecked on the streets. What Iran needs is a humane, practical, and modern solution: 1. A nationwide Catch–Neuter–Vaccinate–Release (CNVR) program 2. Properly managed municipal shelters with medical standards 3. Public education and cultural change We have many urgente matters to attend to after this vile and evil regime falls.

Arsalan Naamdar

33,153 görüntüleme • 7 ay önce

Batch Normalization by hand ✍️ ~ 7 steps walkthrough below Batch normalization is common practice for improving training and achieving faster convergence. It sounds simple. But it is often misunderstood. 🤔 Does batch normalization involve trainable parameters, tunable hyper-parameters, or both? 🤔 Is batch normalization applied to inputs, features, weights, biases, or outputs? 🤔 How is batch normalization different from layer normalization? So I drew and calculated one entirely by hand. Goal: normalize a mini-batch of 4 examples to mean 0 and variance 1, then let the network scale it back. = 1. Given = A mini-batch of 4 training examples, each with 3 features. = 2. Linear layer = Let us multiply by the weights and add the biases. Batch norm sits after this, which answers the second question: what gets normalized is features, not inputs, weights or biases. = 3. ReLU = We apply the activation, and -2 becomes 0. Negative values are suppressed before any statistic is taken. = 4. Batch statistics = Let us compute the sum, mean, variance and standard deviation, one row at a time. A row is a feature and the four columns are the four examples, so every number here measures one feature against the rest of the batch. That is the "batch" in batch normalization, and it is exactly what layer normalization does not do. The statistics are rounded to whole numbers, which is what keeps the rest of the page doable in pen. = 5. Shift to mean 0 = We subtract the mean, in green. The four values in each feature now average to zero. = 6. Scale to variance 1 = Let us divide by the standard deviation, in orange. Each feature now has variance one, whatever scale it arrived at. = 7. Scale and shift = We multiply by a linear transformation and pass the result on. The diagonal and the last column are trainable, so having just forced every feature to mean 0 and variance 1, we hand the network the means to undo it. The outputs: Mean of each feature = [2, 1, 2] Std dev of each feature = [1, 1, 2] To the next layer = [2, -2, 2, 0], [-3, 3, 6, -3], [2, 0, 1, 2] The answers: 🤔 Both. The scale and shift are trainable, the statistics are not. Epsilon and the momentum on the running statistics are the hyper-parameters, and one mini-batch by hand needs neither. 🤔 Features, after the linear layer, not inputs, weights or biases. 🤔 Batch norm measures across the batch, one feature at a time. Layer norm measures across the features, one example at a time. 💾 Save this post!

Batch Normalization by hand ✍️ ~ 7 steps walkthrough below Batch normalization is common practice for improving training and achieving faster convergence. It sounds simple. But it is often misunderstood. 🤔 Does batch normalization involve trainable parameters, tunable hyper-parameters, or both? 🤔 Is batch normalization applied to inputs, features, weights, biases, or outputs? 🤔 How is batch normalization different from layer normalization? So I drew and calculated one entirely by hand. Goal: normalize a mini-batch of 4 examples to mean 0 and variance 1, then let the network scale it back. = 1. Given = A mini-batch of 4 training examples, each with 3 features. = 2. Linear layer = Let us multiply by the weights and add the biases. Batch norm sits after this, which answers the second question: what gets normalized is features, not inputs, weights or biases. = 3. ReLU = We apply the activation, and -2 becomes 0. Negative values are suppressed before any statistic is taken. = 4. Batch statistics = Let us compute the sum, mean, variance and standard deviation, one row at a time. A row is a feature and the four columns are the four examples, so every number here measures one feature against the rest of the batch. That is the "batch" in batch normalization, and it is exactly what layer normalization does not do. The statistics are rounded to whole numbers, which is what keeps the rest of the page doable in pen. = 5. Shift to mean 0 = We subtract the mean, in green. The four values in each feature now average to zero. = 6. Scale to variance 1 = Let us divide by the standard deviation, in orange. Each feature now has variance one, whatever scale it arrived at. = 7. Scale and shift = We multiply by a linear transformation and pass the result on. The diagonal and the last column are trainable, so having just forced every feature to mean 0 and variance 1, we hand the network the means to undo it. The outputs: Mean of each feature = [2, 1, 2] Std dev of each feature = [1, 1, 2] To the next layer = [2, -2, 2, 0], [-3, 3, 6, -3], [2, 0, 1, 2] The answers: 🤔 Both. The scale and shift are trainable, the statistics are not. Epsilon and the momentum on the running statistics are the hyper-parameters, and one mini-batch by hand needs neither. 🤔 Features, after the linear layer, not inputs, weights or biases. 🤔 Batch norm measures across the batch, one feature at a time. Layer norm measures across the features, one example at a time. 💾 Save this post!

Tom Yeh

20,848 görüntüleme • 1 ay önce

Facial reconstruction of a 4,700-year-old Ural Yamnaya/Tamarutkul woman from the Orenburg region The Cis-Ural variant of the Yamnaya culture, also known as the Ilek group, is distinguished by a high proportion of right-side burials and abundant copper tools (Smirnov, 1965; Morgunova & Kravtsov, 1994; Kuznetsov, 2010). The Tamarutkul type was subsequently identified within this group (Bogdanov, 2004). Its origins remain debated: Morgunova (2011; 2014a) considered it a local development rooted in the Volga-Ural Yamnaya and Khvalynsk populations, while others emphasized links with the Poltavka population and a North Caucasian, particularly Novotitorovka, influence (Kuznetsov, 1996a, 2010). Bogdanov (2004) instead proposed influence in the opposite direction. Tamarutkul skulls differed from the contemporaneous Early Poltavka populations, being predominantly mesomorphic, dolichocranial, and strongly profiled, broadly corresponding to the Southeuropid type (Khokhlov, 2017). Similar traits in the Cis-Urals and northwestern Black Sea region have been hypothesized to reflect the spread of related Novotitorovka/"Wagon Burial" groups from the North Caucasus (Khokhlov, 2000, 2013a; Khokhlov et al., 2024). The Novotitorovka culture emerged within the broader Yamnaya cultural sphere, it morphologically resembles the preceding Maykop-Novosvobodnaya culture (Shevchenko, 1983, p. 84; Gei, 2000), likely due to the assimilation of the aforementioned cultural group. The reconstructed individual was buried in a mound measuring 17 m in diameter and 0.58 m in height. One primary Yamnaya burial was excavated beneath the mound. Burial No. 1. A rectangular grave pit (1.9 × 1.2 m) was identified in the southern half of the mound. Its vertical walls reached a depth of 172 cm, and the floor was covered with a thin layer of ochre on which two skeletons lay. Skeleton No. 1 lay crouched on its right side, with the head oriented eastward. The right hand rested beneath the lower jaw, while the left hand rested on the right elbow. All bones were stained with ochre. Skeleton No. 2 (the reconstructed individual) was undisturbed, but all limb bones except the right tibia were absent. Its skull touched the heel bones of Skeleton No. 1 and showed no ochre staining. Near the thigh of Skeleton No. 1 was an oxidized copper object, while a perforated, worn bone object lay near the skull. A ceramic vessel with a thickened base, wide mouth, and outward-flaring rim was found between the pelvises. The surface of the vessel was partially covered with scratches. Ocher powder was preserved inside. The texture of the ceramic is dense, without visible impurities. A copper socketed axe was found in the northwestern corner of the grave. (Morgunova & Kravtsov, 1994) She had a cranial length, cranial width, bizygomatic width, bigonial width, naso-malar angle, and zygo-maxillary angle of 190 mm, 140 mm, 119 mm, 89 mm, 131°, and 124°, respectively.

Facial reconstruction of a 4,700-year-old Ural Yamnaya/Tamarutkul woman from the Orenburg region The Cis-Ural variant of the Yamnaya culture, also known as the Ilek group, is distinguished by a high proportion of right-side burials and abundant copper tools (Smirnov, 1965; Morgunova & Kravtsov, 1994; Kuznetsov, 2010). The Tamarutkul type was subsequently identified within this group (Bogdanov, 2004). Its origins remain debated: Morgunova (2011; 2014a) considered it a local development rooted in the Volga-Ural Yamnaya and Khvalynsk populations, while others emphasized links with the Poltavka population and a North Caucasian, particularly Novotitorovka, influence (Kuznetsov, 1996a, 2010). Bogdanov (2004) instead proposed influence in the opposite direction. Tamarutkul skulls differed from the contemporaneous Early Poltavka populations, being predominantly mesomorphic, dolichocranial, and strongly profiled, broadly corresponding to the Southeuropid type (Khokhlov, 2017). Similar traits in the Cis-Urals and northwestern Black Sea region have been hypothesized to reflect the spread of related Novotitorovka/"Wagon Burial" groups from the North Caucasus (Khokhlov, 2000, 2013a; Khokhlov et al., 2024). The Novotitorovka culture emerged within the broader Yamnaya cultural sphere, it morphologically resembles the preceding Maykop-Novosvobodnaya culture (Shevchenko, 1983, p. 84; Gei, 2000), likely due to the assimilation of the aforementioned cultural group. The reconstructed individual was buried in a mound measuring 17 m in diameter and 0.58 m in height. One primary Yamnaya burial was excavated beneath the mound. Burial No. 1. A rectangular grave pit (1.9 × 1.2 m) was identified in the southern half of the mound. Its vertical walls reached a depth of 172 cm, and the floor was covered with a thin layer of ochre on which two skeletons lay. Skeleton No. 1 lay crouched on its right side, with the head oriented eastward. The right hand rested beneath the lower jaw, while the left hand rested on the right elbow. All bones were stained with ochre. Skeleton No. 2 (the reconstructed individual) was undisturbed, but all limb bones except the right tibia were absent. Its skull touched the heel bones of Skeleton No. 1 and showed no ochre staining. Near the thigh of Skeleton No. 1 was an oxidized copper object, while a perforated, worn bone object lay near the skull. A ceramic vessel with a thickened base, wide mouth, and outward-flaring rim was found between the pelvises. The surface of the vessel was partially covered with scratches. Ocher powder was preserved inside. The texture of the ceramic is dense, without visible impurities. A copper socketed axe was found in the northwestern corner of the grave. (Morgunova & Kravtsov, 1994) She had a cranial length, cranial width, bizygomatic width, bigonial width, naso-malar angle, and zygo-maxillary angle of 190 mm, 140 mm, 119 mm, 89 mm, 131°, and 124°, respectively.

Ancestral Whispers

144,886 görüntüleme • 23 gün önce

(Yes i know the movement is exaggerated, its just to show the motion, it also doeant move like jello when in vr) -Left is before, Right is after- Ok wall of text time for the nerds Literally everything is the same on both sides except the position of 1 bone (made sure that even after adjusting, the collider would still be in the same spot) The reason for this change is because she wanted to be able to clap her ass in vr, and last time i did this all i did was add a toggle for a hidden bone that moved, and its angle would trigger the sound. Which was because that with my previous position, the cheeks could never really meet in the middle when in motion unless you just pushed them together with your hands, and i wanted to just have the sound actually be triggered by them colliding with each other This is also the first time ive used endbones to help drive the movement, so now thanks to that and all my previous experimenting with squishing as well, I can now get bones to move exactly how i want them to move when in natural motion. And i know a lot of people are gonna ask me to teach them how or explain my thought process but i literally just go off feeling lmao. Like, i just kinda visualize in my head how i want the motion to look, and then i just kinda know where to put everything, so im not even sure how to even start explaining rip. I have the things that artist want where they can just make the shit in their head exactly how they envisioned it lol I have a few more ideas i want to try, so we'll see if theyre good enough to get a tweet lol Thank you for coming to my ted talk

(Yes i know the movement is exaggerated, its just to show the motion, it also doeant move like jello when in vr) -Left is before, Right is after- Ok wall of text time for the nerds Literally everything is the same on both sides except the position of 1 bone (made sure that even after adjusting, the collider would still be in the same spot) The reason for this change is because she wanted to be able to clap her ass in vr, and last time i did this all i did was add a toggle for a hidden bone that moved, and its angle would trigger the sound. Which was because that with my previous position, the cheeks could never really meet in the middle when in motion unless you just pushed them together with your hands, and i wanted to just have the sound actually be triggered by them colliding with each other This is also the first time ive used endbones to help drive the movement, so now thanks to that and all my previous experimenting with squishing as well, I can now get bones to move exactly how i want them to move when in natural motion. And i know a lot of people are gonna ask me to teach them how or explain my thought process but i literally just go off feeling lmao. Like, i just kinda visualize in my head how i want the motion to look, and then i just kinda know where to put everything, so im not even sure how to even start explaining rip. I have the things that artist want where they can just make the shit in their head exactly how they envisioned it lol I have a few more ideas i want to try, so we'll see if theyre good enough to get a tweet lol Thank you for coming to my ted talk

Pixel

27,240 görüntüleme • 1 yıl önce

TikTok Phone Farms are the new Gold Rush…. (Apparently) i’m posting over 2500+ TikTok’s this month Here’s the 4 rules I follow whenever I create a fresh TikTok account to guarantee success on day 1. 1. Create the account inside the USA, using a USA based wifi/VPN. 2. Only use 3 TikTok accounts per phone. Any more & TT can flag it as farming. 3. Don’t post on day 1. Warm up the account for 3-7 days. 4. Build an inspo list for your warmup process. Either you, or your team member should engage on the fresh account 2x per day for 5-10 mins. Follow 10-15 accounts from your inspiration list. Like, save, comment and share. Don’t start posting content until your TikTok FYP is only showing content from within your niche. So that you know your account is ‘warm’ and optimized to go viral.

TikTok Phone Farms are the new Gold Rush…. (Apparently) i’m posting over 2500+ TikTok’s this month Here’s the 4 rules I follow whenever I create a fresh TikTok account to guarantee success on day 1. 1. Create the account inside the USA, using a USA based wifi/VPN. 2. Only use 3 TikTok accounts per phone. Any more & TT can flag it as farming. 3. Don’t post on day 1. Warm up the account for 3-7 days. 4. Build an inspo list for your warmup process. Either you, or your team member should engage on the fresh account 2x per day for 5-10 mins. Follow 10-15 accounts from your inspiration list. Like, save, comment and share. Don’t start posting content until your TikTok FYP is only showing content from within your niche. So that you know your account is ‘warm’ and optimized to go viral.

Lachy_eddy | Creative Strategist

186,222 görüntüleme • 1 yıl önce

yesterday, i stumbled onto the most underrated market research tool. tiktok creator insights. it's a goldmine of consumer behavior data, hiding in plain sight. and it's free to use. here's why it's powerful: 1. shows you what people are desperately searching for 2. highlights topics with high demand but low supply 3. reveals trending questions in every industry 4. tracks search growth over 14-day periods the "content gap" tab shows you problems people are actively trying to solve, but can't find good solutions for. so that's cool for a couple reasons 1. help you create content that has low supply/high demand (better chances of going viral) 2. you can build startups to some of these trends Example: i searched "email management" and found: • "how to clear 10k emails" • "best way to organize work inbox" • "email templates for busy people" thousands searching. hardly any solutions. the beauty of this • it's real-time market research • it's actual user intent • it's completely free • and most founders aren't using it a bunch of smart founders are mining tiktok insights right now it isn't perfect, but you never know what you might find your next startup idea might be hiding in those search trends. So, ill share how to access it because it’s kinda hidden: 1. Go to TT search 2.Type in “creator search insight” 3. Tap view im one of those people that think using data like this is your unfair advantage. if tiktok is the new search engine, then tiktok creator insights is the new google trends. might as well use it.

yesterday, i stumbled onto the most underrated market research tool. tiktok creator insights. it's a goldmine of consumer behavior data, hiding in plain sight. and it's free to use. here's why it's powerful: 1. shows you what people are desperately searching for 2. highlights topics with high demand but low supply 3. reveals trending questions in every industry 4. tracks search growth over 14-day periods the "content gap" tab shows you problems people are actively trying to solve, but can't find good solutions for. so that's cool for a couple reasons 1. help you create content that has low supply/high demand (better chances of going viral) 2. you can build startups to some of these trends Example: i searched "email management" and found: • "how to clear 10k emails" • "best way to organize work inbox" • "email templates for busy people" thousands searching. hardly any solutions. the beauty of this • it's real-time market research • it's actual user intent • it's completely free • and most founders aren't using it a bunch of smart founders are mining tiktok insights right now it isn't perfect, but you never know what you might find your next startup idea might be hiding in those search trends. So, ill share how to access it because it’s kinda hidden: 1. Go to TT search 2.Type in “creator search insight” 3. Tap view im one of those people that think using data like this is your unfair advantage. if tiktok is the new search engine, then tiktok creator insights is the new google trends. might as well use it.

GREG ISENBERG

265,916 görüntüleme • 1 yıl önce

$[Backpropagation] by Hand✍️ [1] Forward Pass ↳ Given a multi layer perceptron (3 levels), an input vector X, predictions Y^{Pred} = [0.5, 0.5, 0], and ground truth label Y^{Target} = [0, 1, 0]. [2] Backpropagation ↳ Insert cells to hold our calculations. [3] Layer 3 - Softmax (blue) ↳ Calculate ∂L / ∂z3 directly using the simple equation: Y^{Pred} - Y^{Target} = [0.5, -0.5, 0]. ↳ This simple equation is the benefit of using Softmax and Cross Entropy Loss together. [4] Layer 3 - Weights (orange) & Biases (black) ↳ Calculate ∂L / ∂W3 and ∂L / ∂b3 by multiplying ∂L / ∂z3 and [ a2 | 1 ]. [5] Layer 2 - Activations (green) ↳ Calculate ∂L / ∂a2 by multiplying ∂L / ∂z3 and W3. [6] Layer 2 - ReLU (blue) ↳ Calculate ∂L / ∂z2 by multiplying ∂L / ∂a2 with 1 for positive values and 0 otherwise. [7] Layer 2 - Weights (orange) & Biases (black) ↳ Calculate ∂L / ∂W2 and ∂L / ∂b2 by multiplying ∂L / ∂z2 and [ a1 | 1 ]. [8] Layer 1 - Activations (green) ↳ Calculate ∂L / ∂a1 by multiplying ∂L / ∂z2 and W2. [9] Layer 1 - ReLU (blue) ↳ Calculate ∂L / ∂z1 by multiplying ∂L / ∂a1 with 1 for positive values and 0 otherwise. [10] Layer 1 - Weights (orange) & Biases (black) ↳ Calculate ∂L / ∂W1 and ∂L / ∂b1 by multiplying ∂L / ∂z1 and [ x | 1 ]. [11] Gradient Descent ↳ Update weights and biases (typically a learning rate is applied here). 💡 Matrix Multiplication is All You Need: Just like in the forward pass, backpropagation is all about matrix multiplications. You can definitely do everything by hand as I demonstrated in this exercise, albeit slow and imperfect. This is why GPU's ability to multiply matrices efficiently plays such an important role in the deep learning evolution. This is why NVIDIA is now close to $1 trillion in valuation. 💡Exploding Gradients: We can already see the gradients are getting larger as we back-propagate up, even in this simple 3-layer network. This motivates using methods like skip connections to handle exploding (or diminishing) gradients as in the ResNet. I did the calculations entirely by hand. Please let me know if you spot any error or have any questions!$

[Backpropagation] by Hand✍️ [1] Forward Pass ↳ Given a multi layer perceptron (3 levels), an input vector X, predictions Y^{Pred} = [0.5, 0.5, 0], and ground truth label Y^{Target} = [0, 1, 0]. [2] Backpropagation ↳ Insert cells to hold our calculations. [3] Layer 3 - Softmax (blue) ↳ Calculate ∂L / ∂z3 directly using the simple equation: Y^{Pred} - Y^{Target} = [0.5, -0.5, 0]. ↳ This simple equation is the benefit of using Softmax and Cross Entropy Loss together. [4] Layer 3 - Weights (orange) & Biases (black) ↳ Calculate ∂L / ∂W3 and ∂L / ∂b3 by multiplying ∂L / ∂z3 and [ a2 | 1 ]. [5] Layer 2 - Activations (green) ↳ Calculate ∂L / ∂a2 by multiplying ∂L / ∂z3 and W3. [6] Layer 2 - ReLU (blue) ↳ Calculate ∂L / ∂z2 by multiplying ∂L / ∂a2 with 1 for positive values and 0 otherwise. [7] Layer 2 - Weights (orange) & Biases (black) ↳ Calculate ∂L / ∂W2 and ∂L / ∂b2 by multiplying ∂L / ∂z2 and [ a1 | 1 ]. [8] Layer 1 - Activations (green) ↳ Calculate ∂L / ∂a1 by multiplying ∂L / ∂z2 and W2. [9] Layer 1 - ReLU (blue) ↳ Calculate ∂L / ∂z1 by multiplying ∂L / ∂a1 with 1 for positive values and 0 otherwise. [10] Layer 1 - Weights (orange) & Biases (black) ↳ Calculate ∂L / ∂W1 and ∂L / ∂b1 by multiplying ∂L / ∂z1 and [ x | 1 ]. [11] Gradient Descent ↳ Update weights and biases (typically a learning rate is applied here). 💡 Matrix Multiplication is All You Need: Just like in the forward pass, backpropagation is all about matrix multiplications. You can definitely do everything by hand as I demonstrated in this exercise, albeit slow and imperfect. This is why GPU's ability to multiply matrices efficiently plays such an important role in the deep learning evolution. This is why NVIDIA is now close to $1 trillion in valuation. 💡Exploding Gradients: We can already see the gradients are getting larger as we back-propagate up, even in this simple 3-layer network. This motivates using methods like skip connections to handle exploding (or diminishing) gradients as in the ResNet. I did the calculations entirely by hand. Please let me know if you spot any error or have any questions!

Tom Yeh

64,645 görüntüleme • 2 yıl önce

[VAE] by Hand ✍️ A Variational Auto Encoder (VAE) learns the structure (mean and variance) of hidden features and generates new data from the learned structure. In contrast, GANs only learn to generate new data to fool a discriminator; they may not necessarily know the underlying structure of the data. The International Conference on Learning Representations (ICLR) this year announced its first ever "Test of Time Award" to recognizes the VAE paper, published 10 years ago. This exercise demonstrates how to calculate a VAE by hand. [1] Given: ↳ Three training examples X1, X2, X3 ↳ Copy training examples to the bottom ↳ The purpose is to train the network to reconstruct the training examples. ↳ Since each target is a training example itself, we use the Greek word "auto" which means "self." This crucial step is what makes an autoencoder "auto." [2] Encoder: Layer 1 + ReLU ↳ Multiply inputs with weights and biases ↳ Apply ReLU, crossing out negative values (-1 -> 0) [3] Encoder: Mean and Variance ↳ Multiply features with two sets of weights and biases ↳ 🟩 The first set predicts the means (𝜇) of latent distributions ↳ 🟪 The second set predicts the standard deviation (𝜎) of latent distributions [4] Reparameterization Trick: Random Offset ↳ Sample epsilon ε from the normal distribution with mean = 0 and variance = 1. ↳ The purpose is to randomly pick a offset away from the mean. ↳ Multiply the standard deviation values with epsilon values. ↳ The purpose is to scale the offset by the standard deviation. [5] Reparameterization Trick: Mean + Offset ↳ Add the sampled offset to predicted mean ↳ The result are new parameters or features 🟨 as inputs to the Decoder. [6] Decoder: Layer 1 + ReLU ↳ Multiply input features with weights and biases ↳ Apply ReLU, crossing out negative values. Here, -4 is crossed out. [7] Decoder: Layer 2 ↳ Multiply features with weights and biases ↳ The output is Decoder's attempt to reconstruct the input data X from reparameterized distributions described by 𝜇 and 𝜎. [8]-[10] KL Divergence Loss [8] Loss Gradient: Mean 𝜇 ↳ We want 𝜇 to approach 0. ↳ A lot of math called SGVB simplifies the calculation of loss gradients to simply 𝜇 [9,10] Loss Gradient: Stdev 𝜎 ↳ We want 𝜎 to approach 1. ↳ A lot of math simplifies the calculation to 𝜎 - (1/ 𝜎) [11] Reconstruction Loss ↳ We want the reconstructed data Y (dark 🟧) to be the same as the input data X. ↳ Some math involving Mean Square Error simplifies the calculation to Y - X.

[VAE] by Hand ✍️ A Variational Auto Encoder (VAE) learns the structure (mean and variance) of hidden features and generates new data from the learned structure. In contrast, GANs only learn to generate new data to fool a discriminator; they may not necessarily know the underlying structure of the data. The International Conference on Learning Representations (ICLR) this year announced its first ever "Test of Time Award" to recognizes the VAE paper, published 10 years ago. This exercise demonstrates how to calculate a VAE by hand. [1] Given: ↳ Three training examples X1, X2, X3 ↳ Copy training examples to the bottom ↳ The purpose is to train the network to reconstruct the training examples. ↳ Since each target is a training example itself, we use the Greek word "auto" which means "self." This crucial step is what makes an autoencoder "auto." [2] Encoder: Layer 1 + ReLU ↳ Multiply inputs with weights and biases ↳ Apply ReLU, crossing out negative values (-1 -> 0) [3] Encoder: Mean and Variance ↳ Multiply features with two sets of weights and biases ↳ 🟩 The first set predicts the means (𝜇) of latent distributions ↳ 🟪 The second set predicts the standard deviation (𝜎) of latent distributions [4] Reparameterization Trick: Random Offset ↳ Sample epsilon ε from the normal distribution with mean = 0 and variance = 1. ↳ The purpose is to randomly pick a offset away from the mean. ↳ Multiply the standard deviation values with epsilon values. ↳ The purpose is to scale the offset by the standard deviation. [5] Reparameterization Trick: Mean + Offset ↳ Add the sampled offset to predicted mean ↳ The result are new parameters or features 🟨 as inputs to the Decoder. [6] Decoder: Layer 1 + ReLU ↳ Multiply input features with weights and biases ↳ Apply ReLU, crossing out negative values. Here, -4 is crossed out. [7] Decoder: Layer 2 ↳ Multiply features with weights and biases ↳ The output is Decoder's attempt to reconstruct the input data X from reparameterized distributions described by 𝜇 and 𝜎. [8]-[10] KL Divergence Loss [8] Loss Gradient: Mean 𝜇 ↳ We want 𝜇 to approach 0. ↳ A lot of math called SGVB simplifies the calculation of loss gradients to simply 𝜇 [9,10] Loss Gradient: Stdev 𝜎 ↳ We want 𝜎 to approach 1. ↳ A lot of math simplifies the calculation to 𝜎 - (1/ 𝜎) [11] Reconstruction Loss ↳ We want the reconstructed data Y (dark 🟧) to be the same as the input data X. ↳ Some math involving Mean Square Error simplifies the calculation to Y - X.

Tom Yeh

48,475 görüntüleme • 2 yıl önce

Getting closer to launching my #UE5 tool for finding bugs & optimisation opportunities in your levels! It features 92 different searches to help you quickly identify issues, here is a peek at the full list: Optimisation Searches ///////////////////////////////// Static Meshes: 🔸No LODs 🔸Nanite Disabled & No LODs 🔸Edited Lighting Channels 🔸Mobility Mode 🔸Nanite Enabled/Disabled 🔸Simulate Physics Enabled/Disabled 🔸Custom Depth Pass Enabled/Disabled 🔸Overridden Light Map Resolution 🔸Using Shadow 2 Sided 🔸Same Material In Multiple Slots 🔸Vertex Colours Added In Level Skeletal Meshes: 🔸No LODs 🔸No Max Draw Distance Set 🔸Edited Lighting Channels 🔸Cast Shadows Enabled/Disabled 🔸Custom Depth Pass Enabled/Disabled 🔸Same Material In Multiple Slots Audio: 🔸Auto Activate Enabled 🔸Audio Volume Priority Of X VFX: 🔸Auto Activate Enabled 🔸No Max Draw Distance 🔸Cast Shadow Enabled/Disabled Foliage: 🔸Missing LODs 🔸No Max Cull Distance 🔸Edited Lighting Channels 🔸Mobility Mode 🔸Cast Shadows Enabled/Disabled 🔸Nanite Enabled/Disabled 🔸Two Sided Shadows Enabled/Disabled 🔸Same Material In Multiple Slots 🔸Overridden Light Map Resolution Decal: 🔸Low Fade ‘Screen Size’ Value Text Render: 🔸Cast Shadow Enabled Lights: 🔸No Max Draw Distance 🔸Edited Lighting Channels 🔸Cast Shadows Enabled/Disabled 🔸Mobility Mode 🔸Has Lighting Function Material 🔸Has IES Texture Applied Volumes: 🔸Mobility Mode 🔸Volume Priority Of X Blueprints: 🔸List All Blueprints 🔸Replicates Enabled/Disabled 🔸Default Net Update Frequency 🔸Default Net Cull Distance Squared Misc: 🔸Actors with Tags Applied 🔸Has Specific Tag Bug Searches //////////////////// Static Meshes: 🔸Missing Meshes 🔸Missing Materials 🔸Overridden Materials 🔸Not Default Scale 🔸Collision Disabled/Enabled 🔸Cast Shadow Disabled/Enabled 🔸Hidden In Game Enabled/Disabled 🔸Receives Decals Disabled/Enabled 🔸Affects Nav Disabled/Enabled Skeletal Meshes: 🔸Missing Skeletal Mesh 🔸Missing Materials 🔸Overridden Materials 🔸No Anims Applied 🔸Not Default Scale 🔸Collision Disabled/Enabled 🔸Hidden In-Game Enabled/Disabled 🔸Receives Decals Disabled/Enabled 🔸Affects Nav Enabled/Disabled Audio: 🔸Missing Audio Asset 🔸Has Sound Class Override 🔸Has Volume Override 🔸Attenuation Overrides 🔸Disabled / Missing Reverb Asset VFX: 🔸Missing Asset 🔸Not Default Scale Foliage: 🔸Missing Meshes 🔸Missing Materials 🔸Collision Enabled/Disabled Decals: 🔸Missing Materials 🔸Not Scaled By ‘Decal Size’ 🔸Is Hidden In-game 🔸Decals That Are Temporary / Permanent Text Render: 🔸Using Default / Empty Text 🔸Using Scale Not World Size 🔸Is Hidden In-Game Lights: 🔸Affects World Disabled 🔸Visibility Set To False Volumes: 🔸Collision Is Disabled 🔸Generate Overlap Events Disabled 🔸Set To Block Pawn 🔸Set To Overlap With Pawn 🔸Set To Block Camera 🔸Not Scaled By ‘Brush Size’ 🔸Post Process Blend Radius Less Than X Blueprints: 🔸Is Hidden In-Game If it sounds useful to you, sign up for my tool newsletter to be notified when I release it: #unrealengine | #gamedev | #indiedev

Getting closer to launching my #UE5 tool for finding bugs & optimisation opportunities in your levels! It features 92 different searches to help you quickly identify issues, here is a peek at the full list: Optimisation Searches ///////////////////////////////// Static Meshes: 🔸No LODs 🔸Nanite Disabled & No LODs 🔸Edited Lighting Channels 🔸Mobility Mode 🔸Nanite Enabled/Disabled 🔸Simulate Physics Enabled/Disabled 🔸Custom Depth Pass Enabled/Disabled 🔸Overridden Light Map Resolution 🔸Using Shadow 2 Sided 🔸Same Material In Multiple Slots 🔸Vertex Colours Added In Level Skeletal Meshes: 🔸No LODs 🔸No Max Draw Distance Set 🔸Edited Lighting Channels 🔸Cast Shadows Enabled/Disabled 🔸Custom Depth Pass Enabled/Disabled 🔸Same Material In Multiple Slots Audio: 🔸Auto Activate Enabled 🔸Audio Volume Priority Of X VFX: 🔸Auto Activate Enabled 🔸No Max Draw Distance 🔸Cast Shadow Enabled/Disabled Foliage: 🔸Missing LODs 🔸No Max Cull Distance 🔸Edited Lighting Channels 🔸Mobility Mode 🔸Cast Shadows Enabled/Disabled 🔸Nanite Enabled/Disabled 🔸Two Sided Shadows Enabled/Disabled 🔸Same Material In Multiple Slots 🔸Overridden Light Map Resolution Decal: 🔸Low Fade ‘Screen Size’ Value Text Render: 🔸Cast Shadow Enabled Lights: 🔸No Max Draw Distance 🔸Edited Lighting Channels 🔸Cast Shadows Enabled/Disabled 🔸Mobility Mode 🔸Has Lighting Function Material 🔸Has IES Texture Applied Volumes: 🔸Mobility Mode 🔸Volume Priority Of X Blueprints: 🔸List All Blueprints 🔸Replicates Enabled/Disabled 🔸Default Net Update Frequency 🔸Default Net Cull Distance Squared Misc: 🔸Actors with Tags Applied 🔸Has Specific Tag Bug Searches //////////////////// Static Meshes: 🔸Missing Meshes 🔸Missing Materials 🔸Overridden Materials 🔸Not Default Scale 🔸Collision Disabled/Enabled 🔸Cast Shadow Disabled/Enabled 🔸Hidden In Game Enabled/Disabled 🔸Receives Decals Disabled/Enabled 🔸Affects Nav Disabled/Enabled Skeletal Meshes: 🔸Missing Skeletal Mesh 🔸Missing Materials 🔸Overridden Materials 🔸No Anims Applied 🔸Not Default Scale 🔸Collision Disabled/Enabled 🔸Hidden In-Game Enabled/Disabled 🔸Receives Decals Disabled/Enabled 🔸Affects Nav Enabled/Disabled Audio: 🔸Missing Audio Asset 🔸Has Sound Class Override 🔸Has Volume Override 🔸Attenuation Overrides 🔸Disabled / Missing Reverb Asset VFX: 🔸Missing Asset 🔸Not Default Scale Foliage: 🔸Missing Meshes 🔸Missing Materials 🔸Collision Enabled/Disabled Decals: 🔸Missing Materials 🔸Not Scaled By ‘Decal Size’ 🔸Is Hidden In-game 🔸Decals That Are Temporary / Permanent Text Render: 🔸Using Default / Empty Text 🔸Using Scale Not World Size 🔸Is Hidden In-Game Lights: 🔸Affects World Disabled 🔸Visibility Set To False Volumes: 🔸Collision Is Disabled 🔸Generate Overlap Events Disabled 🔸Set To Block Pawn 🔸Set To Overlap With Pawn 🔸Set To Block Camera 🔸Not Scaled By ‘Brush Size’ 🔸Post Process Blend Radius Less Than X Blueprints: 🔸Is Hidden In-Game If it sounds useful to you, sign up for my tool newsletter to be notified when I release it: #unrealengine | #gamedev | #indiedev

James Rowbotham

21,213 görüntüleme • 1 yıl önce

Boom! Grok Tasks Make It One Of The Most POWERFUL Real-Time AI Systems In The World. — My How to Use Grok Tasks With Hidden Tools For Powerful Daily Output. Grok Tasks are customizable AI workflows that integrate a variety of tools to streamline daily activities, from research and analysis to creative planning and problem-solving. I have been using them for quite sometime and because of the vital heartbeat of news and first person data on X, it is the most powerful AI platform available. By combining Tasks with tools like web searches, X platform interactions, code execution, and media viewers, you can build efficient, automated processes. These tasks work by prompting Grok with a clear description of what you want to achieve, and Grok will intelligently call the necessary tools in sequence or parallel to deliver results. Here's a step-by-step guide to creating and using Grok Tasks: Step 1: Define Your Task Start by clearly outlining the daily activity or goal. Consider what inputs you have (e.g., a URL, a query, or an attachment) and what output you need (e.g., a summary, calculation, or visual analysis). Break it down into subtasks to identify tool needs. For example, if your task involves researching current events, note that you'll need search and browsing capabilities. Step 2: Review Available Tools Familiarize yourself with the tools Grok can access. Here's a quick overview: - Code Execution: Run Python code for calculations, data processing, or simulations using libraries like numpy, pandas, or sympy. - Browse Page: Fetch and summarize content from any website URL with custom instructions. - Web Search: Perform general internet searches, returning results with optional operators like site:. - Web Search With Snippets: Get quick, detailed excerpts from search results for fact-checking. - X Keyword Search: Advanced search for X posts using operators like from:, since:, or filter:. - X Semantic Search: Find semantically related X posts based on a query, with filters for dates or users. - X User Search: Locate X users by name or handle. - X Thread Fetch: Retrieve a full X post thread, including context like replies and parents. - View Image: Analyze an image from a URL or conversation ID. - View X Video: Extract frames and subtitles from an X-hosted video. - Search PDF Attachment: Query a PDF file for relevant pages using keyword or regex modes. - Browse PDF Attachment: View specific pages of a PDF with text and screenshots. Select tools that align with your task. Aim for a mix to handle data gathering, processing, and visualization. Step 3: Craft Your Prompt Write a detailed prompt to Grok describing the task. Include: - The overall goal. - Specific steps or subtasks. - References to tools if you want to guide the process (e.g., "Use web_search to find sources, then code_execution to analyze data"). - Any constraints, like dates or limits. Example prompt: "Create a Grok Task for my morning routine: Search recent X posts about tech news using x_keyword_search, fetch a key thread with x_thread_fetch, and summarize with browse_page on linked articles." Step 4: Submit and Interact Send your prompt to Grok. It will process the task by calling tools as needed, often in parallel for efficiency. Review the output and refine with follow-up prompts if required (e.g., "Expand on that using view_image for visuals"). Iterate to fine-tune the workflow for reuse. Step 5: Save and Reuse Once refined, note the prompt as a template for future use. You can adapt it for similar tasks, making Grok Tasks a habitual part of your day. Finding Grok Tasks To discover existing Grok Tasks or inspiration for new ones, use X searches with tools like x_keyword_search or x_semantic_search (e.g., query: "Grok Tasks examples" with mode: Latest). Browse community-shared threads via x_thread_fetch, or web_search for tutorials on xAI features. Prompt Grok directly: "Show me popular Grok Tasks for productivity." 1 of 3

Boom! Grok Tasks Make It One Of The Most POWERFUL Real-Time AI Systems In The World. — My How to Use Grok Tasks With Hidden Tools For Powerful Daily Output. Grok Tasks are customizable AI workflows that integrate a variety of tools to streamline daily activities, from research and analysis to creative planning and problem-solving. I have been using them for quite sometime and because of the vital heartbeat of news and first person data on X, it is the most powerful AI platform available. By combining Tasks with tools like web searches, X platform interactions, code execution, and media viewers, you can build efficient, automated processes. These tasks work by prompting Grok with a clear description of what you want to achieve, and Grok will intelligently call the necessary tools in sequence or parallel to deliver results. Here's a step-by-step guide to creating and using Grok Tasks: Step 1: Define Your Task Start by clearly outlining the daily activity or goal. Consider what inputs you have (e.g., a URL, a query, or an attachment) and what output you need (e.g., a summary, calculation, or visual analysis). Break it down into subtasks to identify tool needs. For example, if your task involves researching current events, note that you'll need search and browsing capabilities. Step 2: Review Available Tools Familiarize yourself with the tools Grok can access. Here's a quick overview: - Code Execution: Run Python code for calculations, data processing, or simulations using libraries like numpy, pandas, or sympy. - Browse Page: Fetch and summarize content from any website URL with custom instructions. - Web Search: Perform general internet searches, returning results with optional operators like site:. - Web Search With Snippets: Get quick, detailed excerpts from search results for fact-checking. - X Keyword Search: Advanced search for X posts using operators like from:, since:, or filter:. - X Semantic Search: Find semantically related X posts based on a query, with filters for dates or users. - X User Search: Locate X users by name or handle. - X Thread Fetch: Retrieve a full X post thread, including context like replies and parents. - View Image: Analyze an image from a URL or conversation ID. - View X Video: Extract frames and subtitles from an X-hosted video. - Search PDF Attachment: Query a PDF file for relevant pages using keyword or regex modes. - Browse PDF Attachment: View specific pages of a PDF with text and screenshots. Select tools that align with your task. Aim for a mix to handle data gathering, processing, and visualization. Step 3: Craft Your Prompt Write a detailed prompt to Grok describing the task. Include: - The overall goal. - Specific steps or subtasks. - References to tools if you want to guide the process (e.g., "Use web_search to find sources, then code_execution to analyze data"). - Any constraints, like dates or limits. Example prompt: "Create a Grok Task for my morning routine: Search recent X posts about tech news using x_keyword_search, fetch a key thread with x_thread_fetch, and summarize with browse_page on linked articles." Step 4: Submit and Interact Send your prompt to Grok. It will process the task by calling tools as needed, often in parallel for efficiency. Review the output and refine with follow-up prompts if required (e.g., "Expand on that using view_image for visuals"). Iterate to fine-tune the workflow for reuse. Step 5: Save and Reuse Once refined, note the prompt as a template for future use. You can adapt it for similar tasks, making Grok Tasks a habitual part of your day. Finding Grok Tasks To discover existing Grok Tasks or inspiration for new ones, use X searches with tools like x_keyword_search or x_semantic_search (e.g., query: "Grok Tasks examples" with mode: Latest). Browse community-shared threads via x_thread_fetch, or web_search for tutorials on xAI features. Prompt Grok directly: "Show me popular Grok Tasks for productivity." 1 of 3

Brian Roemmele

152,242 görüntüleme • 8 ay önce