正在加载视频...

视频加载失败

返回首页

Switch Transformer by hand ✍️ ~ 13 steps walkthrough below The Switch Transformer, by Fedus, Zoph, and Shazeer in 2022, is one of the papers that made sparse Mixture of Experts practical at scale. Today, frontier models use MoE to pack enormous parameter counts while activating only a small...

ProfTomYeh's profile picture

Tom Yeh

65,785 subscribers

37,410 次观看 • 1 个月前 •via X (Twitter)

LIVE

Anya Rossi• Live Now

Private livecam show

3 条评论

Martin Ronfort 的头像
Martin Ronfort1 个月前

MoE made dense parameter scaling practical. What's remarkable now is seeing active parameter counts scale down so far without losing capability.

Vítor - e/acc 的头像
Vítor - e/acc1 个月前

Beautiful

Henry Nguyen 的头像
Henry Nguyen1 个月前

Hands-on breakdowns like this are how MoE finally clicks—sparse routing feels way less magic once you trace the dispatch/combine steps yourself.

相关视频

ResNet by hand ✍️ ~ 10 steps walkthrough below "Deep Residual Learning for Image Recognition" (Kaiming He, CVPR 2016) is among the most cited papers in all of deep learning. Why does it matter so much? It fixed the exploding and vanishing gradients that kept deep networks from being deep, and made thousands of layers possible. How simple was the fix? An identity matrix. Goal: push three input vectors through a residual block, then through a transformer encoder block, filling in every cell yourself. = 1. Given = A mini batch of three input vectors, 3D, and the weights of the layers ahead. = 2. Linear layer = Let us multiply by the weights, add the bias, and apply ReLU so negatives become 0. Three feature vectors out. This is F(X). = 3. Concatenate = Now the trick. Stack an identity matrix beside the second layer's weights, and stack the input vectors under the features. Draw the lines between rows and columns: those are the skip connections. The identity is the residual. = 4. Linear layer + identity = We multiply the two stacked matrices. The identity carries X straight through while the weights transform it, so a single multiplication computes F(X) + X. Apply ReLU and hand it to the next block. Now watch the same trick inside a transformer, first in attention. = 5. Attention = Let us take three input vectors in 2D, compute the attention matrix, and multiply to get attention weighted vectors. = 6. Concatenate = We stack two identities this time, two residuals, which is how you get 1 + 1, and stack the input vectors with the attention weighted ones. = 7. Add = Multiply the stacked matrices. The identity adds attention to its own input, across the columns, which is how positions get combined. And again in the feed forward layer. = 8. First layer = Let us multiply by the feed forward weights and bias, then ReLU. Three feature vectors. = 9. Concatenate = Stack and link exactly as in step 3: the residual again. = 10. Second layer + identity = We multiply, apply ReLU, and pass the result to the next encoder block. This identity adds across the rows, combining features rather than positions. Takeaway: one simple "add" is what made really deep networks possible. 💾 Save this post!
0:27

ResNet by hand ✍️ ~ 10 steps walkthrough below "Deep Residual Learning for Image Recognition" (Kaiming He, CVPR 2016) is among the most cited papers in all of deep learning. Why does it matter so much? It fixed the exploding and vanishing gradients that kept deep networks from being deep, and made thousands of layers possible. How simple was the fix? An identity matrix. Goal: push three input vectors through a residual block, then through a transformer encoder block, filling in every cell yourself. = 1. Given = A mini batch of three input vectors, 3D, and the weights of the layers ahead. = 2. Linear layer = Let us multiply by the weights, add the bias, and apply ReLU so negatives become 0. Three feature vectors out. This is F(X). = 3. Concatenate = Now the trick. Stack an identity matrix beside the second layer's weights, and stack the input vectors under the features. Draw the lines between rows and columns: those are the skip connections. The identity is the residual. = 4. Linear layer + identity = We multiply the two stacked matrices. The identity carries X straight through while the weights transform it, so a single multiplication computes F(X) + X. Apply ReLU and hand it to the next block. Now watch the same trick inside a transformer, first in attention. = 5. Attention = Let us take three input vectors in 2D, compute the attention matrix, and multiply to get attention weighted vectors. = 6. Concatenate = We stack two identities this time, two residuals, which is how you get 1 + 1, and stack the input vectors with the attention weighted ones. = 7. Add = Multiply the stacked matrices. The identity adds attention to its own input, across the columns, which is how positions get combined. And again in the feed forward layer. = 8. First layer = Let us multiply by the feed forward weights and bias, then ReLU. Three feature vectors. = 9. Concatenate = Stack and link exactly as in step 3: the residual again. = 10. Second layer + identity = We multiply, apply ReLU, and pass the result to the next encoder block. This identity adds across the rows, combining features rather than positions. Takeaway: one simple "add" is what made really deep networks possible. 💾 Save this post!

Tom Yeh

18,152 次观看 • 2 个月前

Self Attention by hand ✍️ ~ 9 steps walkthrough below Self-attention is what enables LLMs to understand context. How does it work? So I drew and calculated one entirely by hand. Goal: turn four 6D features into four 3D attention weighted features, filling in every cell yourself. = 1. Given = Four feature vectors, six dimensions each, one per position. = 2. Query, key, value = Let us multiply the features by WQ, WK and WV. Queries, keys and values all come out of the same four features, and that is what the word "self" is doing in self-attention. = 3. Prepare for MatMul = We copy the queries across the top and the transposed keys down the side. Lining the two up is half the work. = 4. MatMul = Let us multiply K transpose by Q. Every cell is the dot product of one key with one query, which we use as a matching score. That works because the dot product is the numerator of cosine similarity: it is how alike two vectors are, before anyone divides by their lengths. = 5. Scale = We divide by the square root of dk, the dimension of a key vector, here 3. Without it the scores grow with the dimension and a 64-wide head would swamp the softmax. To keep the page doable in pen, the drawing approximates dividing by root 3 with halving. = 6. e to the power = Let us raise e to the power of each score. This is the first half of softmax, and the drawing uses 3 in place of e, which is close enough to do in your head. = 7. Sum = We add up each column: 16, 6, 7 and 12. = 8. Normalize = Let us divide every cell by its column sum. That gives the attention weight matrix in yellow, and each of its four columns is now a probability distribution over the four positions. The decimals are nudged as they are rounded, so every column still sums to exactly 1. = 9. MatMul = We multiply the value vectors by those weights. Each output is a blend of all four values, mixed in the proportion the attention matrix just decided, and it goes to the position-wise feed forward network in the next layer: the FFN box at the bottom of the page. The outputs: Attention weights (A), by column = [.2, .6, 0, .2], [.2, .4, .2, .2], [.4, .2, 0, .4], [.1, .7, .1, .1] Attention weighted features (Z) = [8, 2, 6], [8, 4, 4], [16, 4, 2], [4, 2, 7] The takeaway: attention is a weighted average, and everything before step 9 exists to decide the weights. Compare every position with every other, turn the scores into one distribution per position, then blend. 💾 Save this post!
0:22

Self Attention by hand ✍️ ~ 9 steps walkthrough below Self-attention is what enables LLMs to understand context. How does it work? So I drew and calculated one entirely by hand. Goal: turn four 6D features into four 3D attention weighted features, filling in every cell yourself. = 1. Given = Four feature vectors, six dimensions each, one per position. = 2. Query, key, value = Let us multiply the features by WQ, WK and WV. Queries, keys and values all come out of the same four features, and that is what the word "self" is doing in self-attention. = 3. Prepare for MatMul = We copy the queries across the top and the transposed keys down the side. Lining the two up is half the work. = 4. MatMul = Let us multiply K transpose by Q. Every cell is the dot product of one key with one query, which we use as a matching score. That works because the dot product is the numerator of cosine similarity: it is how alike two vectors are, before anyone divides by their lengths. = 5. Scale = We divide by the square root of dk, the dimension of a key vector, here 3. Without it the scores grow with the dimension and a 64-wide head would swamp the softmax. To keep the page doable in pen, the drawing approximates dividing by root 3 with halving. = 6. e to the power = Let us raise e to the power of each score. This is the first half of softmax, and the drawing uses 3 in place of e, which is close enough to do in your head. = 7. Sum = We add up each column: 16, 6, 7 and 12. = 8. Normalize = Let us divide every cell by its column sum. That gives the attention weight matrix in yellow, and each of its four columns is now a probability distribution over the four positions. The decimals are nudged as they are rounded, so every column still sums to exactly 1. = 9. MatMul = We multiply the value vectors by those weights. Each output is a blend of all four values, mixed in the proportion the attention matrix just decided, and it goes to the position-wise feed forward network in the next layer: the FFN box at the bottom of the page. The outputs: Attention weights (A), by column = [.2, .6, 0, .2], [.2, .4, .2, .2], [.4, .2, 0, .4], [.1, .7, .1, .1] Attention weighted features (Z) = [8, 2, 6], [8, 4, 4], [16, 4, 2], [4, 2, 7] The takeaway: attention is a weighted average, and everything before step 9 exists to decide the weights. Compare every position with every other, turn the scores into one distribution per position, then blend. 💾 Save this post!

Tom Yeh

27,292 次观看 • 2 个月前

[Self-Attention] by Hand ✍️ Self-attention is what enables LLMs to understand context. How does it work? This exercise demonstrates how to calculate a 6-3 attention head by hand. Note that if we have two instances of this, we get 6-6 attention (i.e., multi-head attention, n=2). -- 𝗚𝗼𝗮𝗹 -- Transform [6D Features 🟧] to [3D Attention Weighted Features 🟦] -- 𝗪𝗮𝗹𝗸𝘁𝗵𝗿𝗼𝘂𝗴𝗵 -- [1] Given ↳ A set of 4 feature vectors (6-D): x1,x2,x3,x4 [2] Query, Key, Value ↳ Multiply features x's with linear transformation matrices WQ, WK, and WV, to obtain query vectors (q1,q2,q3,q4), key vectors (k1,k2,k3,k4), and value vectors (v1,v2,v3,v4). ↳ "Self" refers to the fact that both queries and keys are derived from the same set of features. [3] 🟪 Prepare for MatMul ↳ Copy query vectors ↳ Copy the transpose of key vectors [4] 🟪 MatMul ↳ Multiply K^T and Q ↳ This is equivalent to taking dot product between every pair of query and key vectors. ↳ The purpose is to use dot product as an estimate of the "matching score" between every key-value pair. ↳ This estimate makes sense because dot product is the numerator of Cosine Similarity between two vectors. [5] 🟨 Scale ↳ Scale each element by the square root of dk, which is the dimension of key vectors (dk=3). ↳ The purpose is to normalize the impact of the dk on matching scores, even if we scale dk to 32, 64, or 128. ↳ To simplify hand calculation, we approximate [ □/sqrt(3) ] with [ floor(□/2) ]. [6] 🟩 Softmax: e^x ↳ Raise e to the power of the number in each cell ↳ To simplify hand calculation, we approximate e^□ with 3^□. [7] 🟩 Softmax: ∑ ↳ Sum across each column [8] 🟩 Softmax: 1 / sum ↳ For each column, divide each element by the column sum ↳ The purpose is normalize each column so that the numbers sum to 1. In other words, each column is a probability distribution of attention, and we have four of them. ↳ The result is the Attention Weight Matrix (A) (yellow) [9] 🟦 MatMul ↳ Multiply the value vectors (Vs) with the Attention Weight Matrix (A) ↳ The results are the attention weighted features Zs. ↳ They are fed to the position-wise feed forward network in the next layer.
0:32

[Self-Attention] by Hand ✍️ Self-attention is what enables LLMs to understand context. How does it work? This exercise demonstrates how to calculate a 6-3 attention head by hand. Note that if we have two instances of this, we get 6-6 attention (i.e., multi-head attention, n=2). -- 𝗚𝗼𝗮𝗹 -- Transform [6D Features 🟧] to [3D Attention Weighted Features 🟦] -- 𝗪𝗮𝗹𝗸𝘁𝗵𝗿𝗼𝘂𝗴𝗵 -- [1] Given ↳ A set of 4 feature vectors (6-D): x1,x2,x3,x4 [2] Query, Key, Value ↳ Multiply features x's with linear transformation matrices WQ, WK, and WV, to obtain query vectors (q1,q2,q3,q4), key vectors (k1,k2,k3,k4), and value vectors (v1,v2,v3,v4). ↳ "Self" refers to the fact that both queries and keys are derived from the same set of features. [3] 🟪 Prepare for MatMul ↳ Copy query vectors ↳ Copy the transpose of key vectors [4] 🟪 MatMul ↳ Multiply K^T and Q ↳ This is equivalent to taking dot product between every pair of query and key vectors. ↳ The purpose is to use dot product as an estimate of the "matching score" between every key-value pair. ↳ This estimate makes sense because dot product is the numerator of Cosine Similarity between two vectors. [5] 🟨 Scale ↳ Scale each element by the square root of dk, which is the dimension of key vectors (dk=3). ↳ The purpose is to normalize the impact of the dk on matching scores, even if we scale dk to 32, 64, or 128. ↳ To simplify hand calculation, we approximate [ □/sqrt(3) ] with [ floor(□/2) ]. [6] 🟩 Softmax: e^x ↳ Raise e to the power of the number in each cell ↳ To simplify hand calculation, we approximate e^□ with 3^□. [7] 🟩 Softmax: ∑ ↳ Sum across each column [8] 🟩 Softmax: 1 / sum ↳ For each column, divide each element by the column sum ↳ The purpose is normalize each column so that the numbers sum to 1. In other words, each column is a probability distribution of attention, and we have four of them. ↳ The result is the Attention Weight Matrix (A) (yellow) [9] 🟦 MatMul ↳ Multiply the value vectors (Vs) with the Attention Weight Matrix (A) ↳ The results are the attention weighted features Zs. ↳ They are fed to the position-wise feed forward network in the next layer.

Tom Yeh

101,225 次观看 • 2 年前

[RLHF] by Hand ✍️ Yesterday, Jan Leike (Jan Leike) announced he is joining #Anthropic to lead their "super-alignment" mission. He is the co-inventor of Reinforcement Learning with Human Feedback (#RLHF). How does RLHF work? [1] Given ↳ Reward Model (RM) ↳ Large Language Model (LLM) ↳ Two (Prompt, Next) Pairs 🟪 TRAIN RM Goal: Learn to give higher rewards to winners [2] Preferences ↳ A human reviews the two pairs and picks a "winner" ↳ (doc is, him) Embeddings ↳ This prompt has never received human feedback directly ↳ [S] is the special start symbol [11] Transformer ↳ Attention (yellow) ↳ Feed Forward (4x2 weight and bias matrix) ↳ Output: 3 "transformed" feature vector, one per position ↳ More details in my previous post 8. Transformer [] [12] Output Probabilities ↳ Apply a linear layer to map each transformed feature vector to a probability distribution over the vocabulary. [13] Sample ↳ Apply the greedy method, which is to pick the word with the highest score ↳ For output 1 and 2, the model accurately predicts the next word ↳ For 3rd output position, the model's predicts "him" [14] Reward Model ↳ The new pair (CEO is, him) is fed to the reward model ↳ The process is same as [3]-[6] ↳ Output: Reward = 3 [15] Loss Gradient ↳ We set the loss as the negative of the reward. ↳ The loss gradient is simply a constant -1. ↳ Run backpropagation and gradient descent to update LLM's weights and biases (red border)
1:01

[RLHF] by Hand ✍️ Yesterday, Jan Leike (Jan Leike) announced he is joining #Anthropic to lead their "super-alignment" mission. He is the co-inventor of Reinforcement Learning with Human Feedback (#RLHF). How does RLHF work? [1] Given ↳ Reward Model (RM) ↳ Large Language Model (LLM) ↳ Two (Prompt, Next) Pairs 🟪 TRAIN RM Goal: Learn to give higher rewards to winners [2] Preferences ↳ A human reviews the two pairs and picks a "winner" ↳ (doc is, him) Embeddings ↳ This prompt has never received human feedback directly ↳ [S] is the special start symbol [11] Transformer ↳ Attention (yellow) ↳ Feed Forward (4x2 weight and bias matrix) ↳ Output: 3 "transformed" feature vector, one per position ↳ More details in my previous post 8. Transformer [] [12] Output Probabilities ↳ Apply a linear layer to map each transformed feature vector to a probability distribution over the vocabulary. [13] Sample ↳ Apply the greedy method, which is to pick the word with the highest score ↳ For output 1 and 2, the model accurately predicts the next word ↳ For 3rd output position, the model's predicts "him" [14] Reward Model ↳ The new pair (CEO is, him) is fed to the reward model ↳ The process is same as [3]-[6] ↳ Output: Reward = 3 [15] Loss Gradient ↳ We set the loss as the negative of the reward. ↳ The loss gradient is simply a constant -1. ↳ Run backpropagation and gradient descent to update LLM's weights and biases (red border)

Tom Yeh

79,916 次观看 • 2 年前

LSTM by hand ✍️ ~ 15 steps walkthrough below Since Hochreiter and Schmidhuber introduced them in 1997, LSTMs were the most effective way to handle long sequences, right up until the Transformer wave. They are a recurrent network: they read one input at a time and carry a memory forward. Lately recurrence is back in fashion (Mamba), because attention does not scale to hundreds of thousands of tokens. So I drew and calculated one entirely by hand. Goal: run an LSTM cell over a sequence of three inputs, filling in every gate and memory cell yourself. 1. Given Three inputs X1, X2, X3, and four weight matrices: forget, input, candidate, and output. 2. Initialize Let us set the previous hidden state h0 and the memory cell C0 to their start values. 3. Linear transform We multiply the four weight matrices by the stack of the current input, the previous hidden state, and a 1. 4. The gates Let us squash three of those results with sigmoid, giving the forget, input, and output gates, each between 0 and 1. 5. Update the memory We forget part of the old memory (C0 times the forget gate) and add the new (the candidate times the input gate). That is the new memory C1. 6. Candidate output Let us apply tanh to the new memory. 7. Update the hidden state We multiply that candidate by the output gate. The result is h1. 8. Process X2 Copy h1 and C1 forward, then repeat the whole cell: linear transform, the gates, update memory to C2, output gate to h2. 9. Process X3 Once more. Copy h2 and C2 forward, repeat, and read off h3, the final hidden state. Now you can show off to your friends that you calculated an LSTM by hand. ✍️😉 💾 Save this post! #AIbyHand #LSTM #DeepLearning
0:29

LSTM by hand ✍️ ~ 15 steps walkthrough below Since Hochreiter and Schmidhuber introduced them in 1997, LSTMs were the most effective way to handle long sequences, right up until the Transformer wave. They are a recurrent network: they read one input at a time and carry a memory forward. Lately recurrence is back in fashion (Mamba), because attention does not scale to hundreds of thousands of tokens. So I drew and calculated one entirely by hand. Goal: run an LSTM cell over a sequence of three inputs, filling in every gate and memory cell yourself. 1. Given Three inputs X1, X2, X3, and four weight matrices: forget, input, candidate, and output. 2. Initialize Let us set the previous hidden state h0 and the memory cell C0 to their start values. 3. Linear transform We multiply the four weight matrices by the stack of the current input, the previous hidden state, and a 1. 4. The gates Let us squash three of those results with sigmoid, giving the forget, input, and output gates, each between 0 and 1. 5. Update the memory We forget part of the old memory (C0 times the forget gate) and add the new (the candidate times the input gate). That is the new memory C1. 6. Candidate output Let us apply tanh to the new memory. 7. Update the hidden state We multiply that candidate by the output gate. The result is h1. 8. Process X2 Copy h1 and C1 forward, then repeat the whole cell: linear transform, the gates, update memory to C2, output gate to h2. 9. Process X3 Once more. Copy h2 and C2 forward, repeat, and read off h3, the final hidden state. Now you can show off to your friends that you calculated an LSTM by hand. ✍️😉 💾 Save this post! #AIbyHand #LSTM #DeepLearning

Tom Yeh

18,144 次观看 • 2 个月前

SORA by Hand ✍️ OpenAI’s #SORA took over the Internet when it was announced earlier this year. The technology behind Sora is the Diffusion Transformer (DiT) developed by William Peebles and Shining Xie. How does DiT work? 𝗚𝗼𝗮𝗹: Generate a video conditioned by a text prompt and a series of diffusion steps [1] Given ↳ Video ↳ Prompt: "sora is sky" ↳ Diffusion step: t = 3 [2] Video → Patches ↳ Divide all pixels in all frames into 4 spacetime patches [3] Visual Encoder: Pixels 🟨 → Latent 🟩 ↳ Multiply the patches with weights and biases, followed by ReLU ↳ The result is a latent feature vector per patch ↳ The purpose is dimension reduction from 4 (2x2x1) to 2 (2x1). ↳ In the paper, the reduction is 196,608 (256x256x3)→ 4096 (32x32x4) [4] ⬛ Add Noise ↳ Sample a noise according to the diffusion time step t. Typically, the larger the t, the smaller the noise. ↳ Add the Sampled Noise to latent features to obtain Noised Latent. ↳ The goal is to purposely add noise to a video and ask the model to guess what that noise is. ↳ This is analogous to training a language model by purposely deleting a word in a sentence and ask the model to guess what the deleted word was. [5-7] 🟪 Conditioning by Adaptive Layer Norm [5] Encode Conditions ↳ Encode "sora is sky" into a text embedding vector [0,1,-1]. ↳ Encode t = 3 to as a binary vector [1,1]. ↳ Concatenate the two vectors in to a 5D column vector. [6] Estimate Scale/Shift ↳ Multiply the combined vector with weights and biases ↳ The goal is to estimate the scale [2,-1] and shift [-1,5]. ↳ Copy the result to (X) and (+) [7] Apply Scale/Sift ↳ Scale the noised latent by [2,-1] ↳ Shifted the scaled noised latent by [-1, 5] ↳ The result is "conditioned" noise latent. [8-10] Transformer [8] Self-Attention ↳ Feed the conditioned noised latent to Query-Key function to obtain a self-attention matrix ↳ Value is omitted for simplicity [9] Attention Pooling ↳ Multiply the conditioned noised latent with the self-attention matrix ↳ The result are attention weighted features [10] Pointwise Feed Forward Network ↳ Multiply the attention weighted features with weights and biases ↳ The result is the Predicted Noise 🏋️‍♂️ 𝗧𝗿𝗮𝗶𝗻 [11] ↳ Calculate MSE loss gradients by taking the different between the Predicted Noise and the Sampled Noise (ground truth). ↳ Use the loss gradients to kick off backpropagation to update all learnable parameters (red borders) ↳ Note the visual encoder and decoder's parameters are frozen (blue borders) 🎨 𝗚𝗲𝗻𝗲𝗿𝗮𝘁𝗲 (𝗦𝗮𝗺𝗽𝗹𝗲) [12] Denoise ↳ Subtract the predicted noise from the noised latent to obtain the noise-free latent [13] Visual Decoder: Latent 🟩 → Pixels 🟨 ↳ Multiply the patches with weights and biases, followed by ReLU [14] Patches → Video ↳ Rearrange patches into a sequence of video frames.
0:43

SORA by Hand ✍️ OpenAI’s #SORA took over the Internet when it was announced earlier this year. The technology behind Sora is the Diffusion Transformer (DiT) developed by William Peebles and Shining Xie. How does DiT work? 𝗚𝗼𝗮𝗹: Generate a video conditioned by a text prompt and a series of diffusion steps [1] Given ↳ Video ↳ Prompt: "sora is sky" ↳ Diffusion step: t = 3 [2] Video → Patches ↳ Divide all pixels in all frames into 4 spacetime patches [3] Visual Encoder: Pixels 🟨 → Latent 🟩 ↳ Multiply the patches with weights and biases, followed by ReLU ↳ The result is a latent feature vector per patch ↳ The purpose is dimension reduction from 4 (2x2x1) to 2 (2x1). ↳ In the paper, the reduction is 196,608 (256x256x3)→ 4096 (32x32x4) [4] ⬛ Add Noise ↳ Sample a noise according to the diffusion time step t. Typically, the larger the t, the smaller the noise. ↳ Add the Sampled Noise to latent features to obtain Noised Latent. ↳ The goal is to purposely add noise to a video and ask the model to guess what that noise is. ↳ This is analogous to training a language model by purposely deleting a word in a sentence and ask the model to guess what the deleted word was. [5-7] 🟪 Conditioning by Adaptive Layer Norm [5] Encode Conditions ↳ Encode "sora is sky" into a text embedding vector [0,1,-1]. ↳ Encode t = 3 to as a binary vector [1,1]. ↳ Concatenate the two vectors in to a 5D column vector. [6] Estimate Scale/Shift ↳ Multiply the combined vector with weights and biases ↳ The goal is to estimate the scale [2,-1] and shift [-1,5]. ↳ Copy the result to (X) and (+) [7] Apply Scale/Sift ↳ Scale the noised latent by [2,-1] ↳ Shifted the scaled noised latent by [-1, 5] ↳ The result is "conditioned" noise latent. [8-10] Transformer [8] Self-Attention ↳ Feed the conditioned noised latent to Query-Key function to obtain a self-attention matrix ↳ Value is omitted for simplicity [9] Attention Pooling ↳ Multiply the conditioned noised latent with the self-attention matrix ↳ The result are attention weighted features [10] Pointwise Feed Forward Network ↳ Multiply the attention weighted features with weights and biases ↳ The result is the Predicted Noise 🏋️‍♂️ 𝗧𝗿𝗮𝗶𝗻 [11] ↳ Calculate MSE loss gradients by taking the different between the Predicted Noise and the Sampled Noise (ground truth). ↳ Use the loss gradients to kick off backpropagation to update all learnable parameters (red borders) ↳ Note the visual encoder and decoder's parameters are frozen (blue borders) 🎨 𝗚𝗲𝗻𝗲𝗿𝗮𝘁𝗲 (𝗦𝗮𝗺𝗽𝗹𝗲) [12] Denoise ↳ Subtract the predicted noise from the noised latent to obtain the noise-free latent [13] Visual Decoder: Latent 🟩 → Pixels 🟨 ↳ Multiply the patches with weights and biases, followed by ReLU [14] Patches → Video ↳ Rearrange patches into a sequence of video frames.

Tom Yeh

238,409 次观看 • 2 年前

Scaled Dot-Product Attention by hand ✍️ I explained attention in my recent AI seminar as follows: Step 1 — Compare We take one token as a query and compare it against all keys using dot products. This gives us a grid of similarity scores: every query against every key. Step 2 — Scale We divide those scores by √dₖ to keep the values numerically stable. Step 3 — Normalize We apply softmax to turn raw scores into a probability distribution—values between 0 and 1 that sum to 1. These are the attention weights. Step 4 — Combine We use those weights to compute a weighted sum of the values. Each output token becomes a linear combination of all previous tokens—maybe 37% from one, 23% from another, and a little from the rest. No blackbox. Just dot products, scaling, softmax, and weighted sums—built one row at a time in Excel.
0:29

Scaled Dot-Product Attention by hand ✍️ I explained attention in my recent AI seminar as follows: Step 1 — Compare We take one token as a query and compare it against all keys using dot products. This gives us a grid of similarity scores: every query against every key. Step 2 — Scale We divide those scores by √dₖ to keep the values numerically stable. Step 3 — Normalize We apply softmax to turn raw scores into a probability distribution—values between 0 and 1 that sum to 1. These are the attention weights. Step 4 — Combine We use those weights to compute a weighted sum of the values. Each output token becomes a linear combination of all previous tokens—maybe 37% from one, 23% from another, and a little from the rest. No blackbox. Just dot products, scaling, softmax, and weighted sums—built one row at a time in Excel.

Tom Yeh

31,114 次观看 • 8 个月前

Graph Convolutional Network by hand ✍️ ~ 12 steps walkthrough below Graph Convolutional Networks (GCNs), introduced by Thomas Kipf and Max Welling in 2017, are the tool for data shaped like a graph: social networks, recommendations, biological networks, drug discovery, molecular chemistry. I drew and calculated a simple GCN entirely by hand. Goal: run a two-layer GCN, then a small classifier, on a five-node graph, filling in every cell yourself. 1. Given A graph of five nodes, A to E, with edges between some of them. 2. Adjacency matrix (neighbors) Put a 1 wherever two nodes share an edge, in both directions. 3. Adjacency matrix (self) Add 1s down the diagonal, one self-loop per node. That is just adding the identity matrix. 4. Messages Multiply each node's embedding by the weights and biases, then ReLU. Negatives become 0. 5. Pooling Multiply the messages by the adjacency matrix. Each node gathers the messages of its neighbours and itself. 6. Visualize Node A pools [3,0,1] + [1,0,0] = [4,0,1]. 7. Second GCN layer Messages again: weights, biases, ReLU. 8. Pooling again Pool over each node and its neighbours, once more. 9. Visualize Node C pools [1,2,4] + [1,3,5] + [0,0,1] = [2,5,10]. 10. Fully connected layer Weights, biases, ReLU. This time there are no neighbours to pool, just the node itself. 11. Linear layer One more: weights and biases. 12. Sigmoid Squash each score to a probability (≥ 3 → 1, 0 → 0.5, ≤ -3 → 0). That is the classification for each node. You have just classified every node in the graph by hand. ✍️ The outputs: A: 0 (very unlikely) B: 1 (very likely) C: 1 (very likely) D: 1 (very likely) E: 0.5 (neutral) The takeaway: a GCN layer is two parts. The top part pools each node with its neighbours through the adjacency matrix. The bottom part is an MLP that transforms each node on its own. A transformer layer has the same two parts, with an attention matrix where the adjacency matrix was. Both matrices do one job, mixing across positions: attention over tokens, adjacency over nodes. In my class I call the GCN the transformer's little cousin: a bit more stubborn, because its attention is fixed by the graph rather than computed from Q, K, and V. Draw the two side by side and the resemblance is hard to miss. 💾 Save this post! #AIbyHand #GraphNeuralNetworks #DeepLearning
0:23

Graph Convolutional Network by hand ✍️ ~ 12 steps walkthrough below Graph Convolutional Networks (GCNs), introduced by Thomas Kipf and Max Welling in 2017, are the tool for data shaped like a graph: social networks, recommendations, biological networks, drug discovery, molecular chemistry. I drew and calculated a simple GCN entirely by hand. Goal: run a two-layer GCN, then a small classifier, on a five-node graph, filling in every cell yourself. 1. Given A graph of five nodes, A to E, with edges between some of them. 2. Adjacency matrix (neighbors) Put a 1 wherever two nodes share an edge, in both directions. 3. Adjacency matrix (self) Add 1s down the diagonal, one self-loop per node. That is just adding the identity matrix. 4. Messages Multiply each node's embedding by the weights and biases, then ReLU. Negatives become 0. 5. Pooling Multiply the messages by the adjacency matrix. Each node gathers the messages of its neighbours and itself. 6. Visualize Node A pools [3,0,1] + [1,0,0] = [4,0,1]. 7. Second GCN layer Messages again: weights, biases, ReLU. 8. Pooling again Pool over each node and its neighbours, once more. 9. Visualize Node C pools [1,2,4] + [1,3,5] + [0,0,1] = [2,5,10]. 10. Fully connected layer Weights, biases, ReLU. This time there are no neighbours to pool, just the node itself. 11. Linear layer One more: weights and biases. 12. Sigmoid Squash each score to a probability (≥ 3 → 1, 0 → 0.5, ≤ -3 → 0). That is the classification for each node. You have just classified every node in the graph by hand. ✍️ The outputs: A: 0 (very unlikely) B: 1 (very likely) C: 1 (very likely) D: 1 (very likely) E: 0.5 (neutral) The takeaway: a GCN layer is two parts. The top part pools each node with its neighbours through the adjacency matrix. The bottom part is an MLP that transforms each node on its own. A transformer layer has the same two parts, with an attention matrix where the adjacency matrix was. Both matrices do one job, mixing across positions: attention over tokens, adjacency over nodes. In my class I call the GCN the transformer's little cousin: a bit more stubborn, because its attention is fixed by the graph rather than computed from Q, K, and V. Draw the two side by side and the resemblance is hard to miss. 💾 Save this post! #AIbyHand #GraphNeuralNetworks #DeepLearning

Tom Yeh

16,800 次观看 • 2 个月前

SVM by hand ✍️ ~ 19 steps walkthrough below (Linear vs RBF) Support Vector Machines reigned supreme in machine learning before the deep learning revolution. An SVM predicts with dot products, the same matrix multiplication every model uses. What it does not do is train by backpropagation: it is fitted by convex optimization, so there is no matrix-multiplication backward pass for a GPU to accelerate. I drew and calculated two SVMs by hand: a linear one (top) and an RBF one (bottom), classifying the same two test vectors. Goal: turn six training vectors and their learned coefficients into a prediction, and see what changing the kernel actually changes. = 1. Given = Six training vectors, their labels, and the coefficients and bias already learned. A coefficient of zero means that vector is not a support vector: too far from the boundary to matter. = 2. Linear kernel, test vector 1 = Let us take the dot product of the test vector with every training vector. The dot product stands in for cosine similarity, and the column of results is the first column of the kernel matrix K. = 3. Linear kernel, test vector 2 = We do the same for the second, and K is complete. = 4. Signed weights = Let us multiply each coefficient by its label. The second training vector drops out here, because its coefficient is 0. = 5. Weighted combination = We multiply the signed weights through K and add the bias b. The result is a signed distance to the decision boundary: 17 and 5. = 6. Classify = Let us take the sign. Both are positive. = 7 to 11. RBF kernel, test vector 1 = Now the same picture with a different kernel, in five moves: square the differences, sum them, take the square root for the L2 distance, multiply by minus gamma, and raise e to that power. The negation is what turns a distance into a similarity, and gamma controls how far a single training vector's influence reaches. = 12 to 16. RBF kernel, test vector 2 = We repeat all five. The numbers change, the moves do not. = 17 to 19. Decision boundary, again = Signed weights, weighted combination, sign. Identical arithmetic to steps 4 through 6, on a K that was built a completely different way. The outputs: Linear K, first column = [13, 25, 12, 15, 19, 27] Linear decision values = 17 and 5, both positive RBF decision values = -2 and 1, so negative and positive The takeaway: the kernel is the only thing that changed, and it changed the answer. The linear SVM calls both test vectors positive; the RBF one splits them. Everything after the kernel matrix, the signed weights and the weighted combination and the sign, is the same page of arithmetic twice. 💾 Save this post!
0:44

SVM by hand ✍️ ~ 19 steps walkthrough below (Linear vs RBF) Support Vector Machines reigned supreme in machine learning before the deep learning revolution. An SVM predicts with dot products, the same matrix multiplication every model uses. What it does not do is train by backpropagation: it is fitted by convex optimization, so there is no matrix-multiplication backward pass for a GPU to accelerate. I drew and calculated two SVMs by hand: a linear one (top) and an RBF one (bottom), classifying the same two test vectors. Goal: turn six training vectors and their learned coefficients into a prediction, and see what changing the kernel actually changes. = 1. Given = Six training vectors, their labels, and the coefficients and bias already learned. A coefficient of zero means that vector is not a support vector: too far from the boundary to matter. = 2. Linear kernel, test vector 1 = Let us take the dot product of the test vector with every training vector. The dot product stands in for cosine similarity, and the column of results is the first column of the kernel matrix K. = 3. Linear kernel, test vector 2 = We do the same for the second, and K is complete. = 4. Signed weights = Let us multiply each coefficient by its label. The second training vector drops out here, because its coefficient is 0. = 5. Weighted combination = We multiply the signed weights through K and add the bias b. The result is a signed distance to the decision boundary: 17 and 5. = 6. Classify = Let us take the sign. Both are positive. = 7 to 11. RBF kernel, test vector 1 = Now the same picture with a different kernel, in five moves: square the differences, sum them, take the square root for the L2 distance, multiply by minus gamma, and raise e to that power. The negation is what turns a distance into a similarity, and gamma controls how far a single training vector's influence reaches. = 12 to 16. RBF kernel, test vector 2 = We repeat all five. The numbers change, the moves do not. = 17 to 19. Decision boundary, again = Signed weights, weighted combination, sign. Identical arithmetic to steps 4 through 6, on a K that was built a completely different way. The outputs: Linear K, first column = [13, 25, 12, 15, 19, 27] Linear decision values = 17 and 5, both positive RBF decision values = -2 and 1, so negative and positive The takeaway: the kernel is the only thing that changed, and it changed the answer. The linear SVM calls both test vectors positive; the RBF one splits them. Everything after the kernel matrix, the signed weights and the weighted combination and the sign, is the same page of arithmetic twice. 💾 Save this post!

Tom Yeh

16,916 次观看 • 2 个月前

Dropout by hand ✍️ ~ 10 steps walkthrough below Dropout is the simplest trick in deep learning that actually works: during training you randomly switch neurons off, so the network cannot lean on any one of them. It is two lines of code and almost nobody has worked through what those lines do to the numbers. So I drew and calculated one entirely by hand. Goal: train one pass through a small network with two dropout layers, then run inference with dropout switched off. The network: Linear(2,4), ReLU, Dropout(0.5), Linear(4,3), ReLU, Dropout(0.33), Linear(3,2). = 1. Given = A training set of two examples, X1 and X2, and the weight matrices for all three linear layers. = 2. Draw the first random numbers = Let us draw 4 random numbers, one per neuron in the first hidden layer. Above 0.5 we keep (◯), below we drop (╳). Here that gives [◯, ╳, ◯, ╳]. = 3. Build the first dropout matrix = We turn that pattern into a diagonal matrix. The scaling factor is 1/(1-p) = 2, so a kept neuron gets 2 and a dropped one gets 0. Multiplying by it does both jobs at once: it deletes the 2nd and 4th neurons and doubles the two that survive. = 4. Draw the second random numbers = Let us do it again for the 3 neurons in the next layer, this time against p = 0.33. The result is [◯, ◯, ╳]. = 5. Build the second dropout matrix = We set the diagonal to 1.5 where kept and 0 where dropped. Only the 3rd neuron goes. = 6. Feed forward = Let us run the whole thing top to bottom: one matrix multiplication per layer, ReLU setting the negatives to zero, and the two dropout matrices doing their work in between. The outputs Y come out at the bottom. = 7. MSE loss gradients = We compare Y against the targets Y', subtract, and multiply each element by 2. That is the whole gradient of the mean squared error. = 8. Update the weights = Let us push those gradients back through the network and update the weights (marked in light red). = 9. Deactivate dropout = Training is over, so we set both dropout matrices to the identity. Every neuron is back, and nothing is scaled. = 10. Feed forward again = One more pass, this time on unseen data, to make the prediction. You have just trained and run a network with dropout by hand. ✍️ The outputs: Training outputs Y = [-6, 9; 13, 4] Loss gradients = [-4, 4; 6, -2] Inference outputs = [13, 13; 4, 3] 💾 Save this post! #AIbyHand #Dropout #DeepLearning #NeuralNetworks
0:25

Dropout by hand ✍️ ~ 10 steps walkthrough below Dropout is the simplest trick in deep learning that actually works: during training you randomly switch neurons off, so the network cannot lean on any one of them. It is two lines of code and almost nobody has worked through what those lines do to the numbers. So I drew and calculated one entirely by hand. Goal: train one pass through a small network with two dropout layers, then run inference with dropout switched off. The network: Linear(2,4), ReLU, Dropout(0.5), Linear(4,3), ReLU, Dropout(0.33), Linear(3,2). = 1. Given = A training set of two examples, X1 and X2, and the weight matrices for all three linear layers. = 2. Draw the first random numbers = Let us draw 4 random numbers, one per neuron in the first hidden layer. Above 0.5 we keep (◯), below we drop (╳). Here that gives [◯, ╳, ◯, ╳]. = 3. Build the first dropout matrix = We turn that pattern into a diagonal matrix. The scaling factor is 1/(1-p) = 2, so a kept neuron gets 2 and a dropped one gets 0. Multiplying by it does both jobs at once: it deletes the 2nd and 4th neurons and doubles the two that survive. = 4. Draw the second random numbers = Let us do it again for the 3 neurons in the next layer, this time against p = 0.33. The result is [◯, ◯, ╳]. = 5. Build the second dropout matrix = We set the diagonal to 1.5 where kept and 0 where dropped. Only the 3rd neuron goes. = 6. Feed forward = Let us run the whole thing top to bottom: one matrix multiplication per layer, ReLU setting the negatives to zero, and the two dropout matrices doing their work in between. The outputs Y come out at the bottom. = 7. MSE loss gradients = We compare Y against the targets Y', subtract, and multiply each element by 2. That is the whole gradient of the mean squared error. = 8. Update the weights = Let us push those gradients back through the network and update the weights (marked in light red). = 9. Deactivate dropout = Training is over, so we set both dropout matrices to the identity. Every neuron is back, and nothing is scaled. = 10. Feed forward again = One more pass, this time on unseen data, to make the prediction. You have just trained and run a network with dropout by hand. ✍️ The outputs: Training outputs Y = [-6, 9; 13, 4] Loss gradients = [-4, 4; 6, -2] Inference outputs = [13, 13; 4, 3] 💾 Save this post! #AIbyHand #Dropout #DeepLearning #NeuralNetworks

Tom Yeh

14,442 次观看 • 2 个月前

The future of AI belongs to the humans behind it. There’s a common fear that as AI gets better, people get pushed out of the picture. We believe the opposite is happening. AI is creating entirely new categories of work, and an entirely new economy around the people whose knowledge, judgment, and experience are helping these systems improve. Today, tens of thousands of experts are actively contributing to AI training projects through micro1. Over time, we believe this will grow to tens of millions+. And if humans are going to play such a critical role in building the future of AI, the companies they work with should raise the standard for how they’re supported. Today, I’m proud to announce the micro1 Expert Support Program. We’re building a new set of protections, resources, and support for our expert community, starting with: -Expert Bill of Rights: a clear set of commitments outlining what experts can expect when working with micro1. -Rest Credits: paid time away from projects when experts need or want a break. -Expert Emergency Fund: financial support for experts facing emergencies. -Mental Health & Coaching: new resources to support expert wellbeing and growth. -Confidential Support Line: a confidential way to ask questions, seek support, or raise concerns related to pipelines. Within the next two weeks, every expert currently active with micro1 will receive an email with the full program details and launch dates. AI is going to keep getting more capable. The opportunity in front of us is to make sure the people helping build it benefit from that progress too.
1:00

The future of AI belongs to the humans behind it. There’s a common fear that as AI gets better, people get pushed out of the picture. We believe the opposite is happening. AI is creating entirely new categories of work, and an entirely new economy around the people whose knowledge, judgment, and experience are helping these systems improve. Today, tens of thousands of experts are actively contributing to AI training projects through micro1. Over time, we believe this will grow to tens of millions+. And if humans are going to play such a critical role in building the future of AI, the companies they work with should raise the standard for how they’re supported. Today, I’m proud to announce the micro1 Expert Support Program. We’re building a new set of protections, resources, and support for our expert community, starting with: -Expert Bill of Rights: a clear set of commitments outlining what experts can expect when working with micro1. -Rest Credits: paid time away from projects when experts need or want a break. -Expert Emergency Fund: financial support for experts facing emergencies. -Mental Health & Coaching: new resources to support expert wellbeing and growth. -Confidential Support Line: a confidential way to ask questions, seek support, or raise concerns related to pipelines. Within the next two weeks, every expert currently active with micro1 will receive an email with the full program details and launch dates. AI is going to keep getting more capable. The opportunity in front of us is to make sure the people helping build it benefit from that progress too.

Ali Ansari

147,250 次观看 • 1 个月前

Discrete Fourier Transform by hand ✍️ ~ 12 steps walkthrough below Here is a little-known secret about the DFT and the inverse DFT: it is just matrix multiplication in both directions, one the transpose of the other, exactly like the forward pass and backpropagation I drew in other examples. Goal: recover which cosine waves a signal is made of, using nothing but multiplication and addition. = 1. Given = Three signals written as sums of cosines, and a fourth, X, that we do not know yet. = 2. Frequency matrix F = Let us write the coefficients as a matrix. Each signal is a row, each frequency a column, so A = cos(w) + 2cos(2w) becomes [1, 2, 0, 0]. = 3. Sample the waves = We read the four cosine waves at ten discrete time points. That word "discrete" is the whole difference between this and the continuous transform. = 4. Cosine matrix W = Let us write those samples as a matrix: each frequency a row, each time point a column. = 5. Frequency to time = We multiply F by W. That combines the four cosine waves in the proportions F specifies, and the result T is the three signals as they would look in time. = 6. Transpose = Let us stand each signal up as a column. = 7. Time to frequency = We multiply W by that transpose. Every cell is the dot product of one signal with one cosine wave, which measures how much of that wave the signal contains. Zero means none of it. = 8. Scale = Let us multiply by 2/n, with n = 10. The projections come out five times too large, and this is the correction. = 9. Transpose back = We turn it back around, and it is F again, exactly. That is the check: the transform recovered the coefficients we started from. = 10. Now solve for X = Let us run the same multiplication on the one signal whose recipe we never knew. = 11. Scale = We divide by 5 again. = 12. Transpose back = And X reads [0, 0, 3, 2], which says X = 3cos(3w) + 2cos(4w). Note: I originally drew this to show that the DFT is a special case of a convolution layer, its filters fixed to sine and cosine waves rather than learned. No wonder, then, that a convolution layer free to learn its own filters can be trained to process signals. 💾 Save this post!
0:32

Discrete Fourier Transform by hand ✍️ ~ 12 steps walkthrough below Here is a little-known secret about the DFT and the inverse DFT: it is just matrix multiplication in both directions, one the transpose of the other, exactly like the forward pass and backpropagation I drew in other examples. Goal: recover which cosine waves a signal is made of, using nothing but multiplication and addition. = 1. Given = Three signals written as sums of cosines, and a fourth, X, that we do not know yet. = 2. Frequency matrix F = Let us write the coefficients as a matrix. Each signal is a row, each frequency a column, so A = cos(w) + 2cos(2w) becomes [1, 2, 0, 0]. = 3. Sample the waves = We read the four cosine waves at ten discrete time points. That word "discrete" is the whole difference between this and the continuous transform. = 4. Cosine matrix W = Let us write those samples as a matrix: each frequency a row, each time point a column. = 5. Frequency to time = We multiply F by W. That combines the four cosine waves in the proportions F specifies, and the result T is the three signals as they would look in time. = 6. Transpose = Let us stand each signal up as a column. = 7. Time to frequency = We multiply W by that transpose. Every cell is the dot product of one signal with one cosine wave, which measures how much of that wave the signal contains. Zero means none of it. = 8. Scale = Let us multiply by 2/n, with n = 10. The projections come out five times too large, and this is the correction. = 9. Transpose back = We turn it back around, and it is F again, exactly. That is the check: the transform recovered the coefficients we started from. = 10. Now solve for X = Let us run the same multiplication on the one signal whose recipe we never knew. = 11. Scale = We divide by 5 again. = 12. Transpose back = And X reads [0, 0, 3, 2], which says X = 3cos(3w) + 2cos(4w). Note: I originally drew this to show that the DFT is a special case of a convolution layer, its filters fixed to sine and cosine waves rather than learned. No wonder, then, that a convolution layer free to learn its own filters can be trained to process signals. 💾 Save this post!

Tom Yeh

13,476 次观看 • 23 天前

Discrete Fourier Transform by hand ✍️ ~ 12 steps walkthrough below Here is a little-known secret about the DFT and the inverse DFT: it is just matrix multiplication in both directions, one the transpose of the other, exactly like the forward pass and backpropagation I drew in other examples. Goal: recover which cosine waves a signal is made of, using nothing but multiplication and addition. = 1. Given = Three signals written as sums of cosines, and a fourth, X, that we do not know yet. = 2. Frequency matrix F = Let us write the coefficients as a matrix. Each signal is a row, each frequency a column, so A = cos(w) + 2cos(2w) becomes [1, 2, 0, 0]. = 3. Sample the waves = We read the four cosine waves at ten discrete time points. That word "discrete" is the whole difference between this and the continuous transform. = 4. Cosine matrix W = Let us write those samples as a matrix: each frequency a row, each time point a column. = 5. Frequency to time = We multiply F by W. That combines the four cosine waves in the proportions F specifies, and the result T is the three signals as they would look in time. = 6. Transpose = Let us stand each signal up as a column. = 7. Time to frequency = We multiply W by that transpose. Every cell is the dot product of one signal with one cosine wave, which measures how much of that wave the signal contains. Zero means none of it. = 8. Scale = Let us multiply by 2/n, with n = 10. The projections come out five times too large, and this is the correction. = 9. Transpose back = We turn it back around, and it is F again, exactly. That is the check: the transform recovered the coefficients we started from. = 10. Now solve for X = Let us run the same multiplication on the one signal whose recipe we never knew. = 11. Scale = We divide by 5 again. = 12. Transpose back = And X reads [0, 0, 3, 2], which says X = 3cos(3w) + 2cos(4w). Note: I originally drew this to show that the DFT is a special case of a convolution layer, its filters fixed to sine and cosine waves rather than learned. No wonder, then, that a convolution layer free to learn its own filters can be trained to process signals. 💾 Save this post!
0:32

Discrete Fourier Transform by hand ✍️ ~ 12 steps walkthrough below Here is a little-known secret about the DFT and the inverse DFT: it is just matrix multiplication in both directions, one the transpose of the other, exactly like the forward pass and backpropagation I drew in other examples. Goal: recover which cosine waves a signal is made of, using nothing but multiplication and addition. = 1. Given = Three signals written as sums of cosines, and a fourth, X, that we do not know yet. = 2. Frequency matrix F = Let us write the coefficients as a matrix. Each signal is a row, each frequency a column, so A = cos(w) + 2cos(2w) becomes [1, 2, 0, 0]. = 3. Sample the waves = We read the four cosine waves at ten discrete time points. That word "discrete" is the whole difference between this and the continuous transform. = 4. Cosine matrix W = Let us write those samples as a matrix: each frequency a row, each time point a column. = 5. Frequency to time = We multiply F by W. That combines the four cosine waves in the proportions F specifies, and the result T is the three signals as they would look in time. = 6. Transpose = Let us stand each signal up as a column. = 7. Time to frequency = We multiply W by that transpose. Every cell is the dot product of one signal with one cosine wave, which measures how much of that wave the signal contains. Zero means none of it. = 8. Scale = Let us multiply by 2/n, with n = 10. The projections come out five times too large, and this is the correction. = 9. Transpose back = We turn it back around, and it is F again, exactly. That is the check: the transform recovered the coefficients we started from. = 10. Now solve for X = Let us run the same multiplication on the one signal whose recipe we never knew. = 11. Scale = We divide by 5 again. = 12. Transpose back = And X reads [0, 0, 3, 2], which says X = 3cos(3w) + 2cos(4w). Note: I originally drew this to show that the DFT is a special case of a convolution layer, its filters fixed to sine and cosine waves rather than learned. No wonder, then, that a convolution layer free to learn its own filters can be trained to process signals. 💾 Save this post!

Tom Yeh

25,684 次观看 • 2 个月前

Variational Autoencoder by hand ✍️ ~ 11 steps walkthrough below A VAE learns the structure of your data, the mean and variance of its hidden features, and then generates new data from that structure. A GAN only learns to fool a discriminator. It can make convincing fakes without ever knowing what the data is really made of. That is the difference, and it is the whole reason VAEs matter. In 2024 ICLR gave its first ever Test of Time Award to the VAE paper, "Auto-Encoding Variational Bayes" by Diederik Kingma and Max Welling, ten years on. How does it work? Goal: encode three inputs into a distribution, sample from it, decode it back, and read every loss gradient off the page. = 1. Given = Three training examples X1, X2, X3, copied to the bottom as their own targets. Reconstructing your own input is what puts the "auto", meaning self, in autoencoder. = 2. Encoder, layer 1 = Let us multiply the inputs by weights and biases, then apply ReLU, crossing out every negative. = 3. Mean and standard deviation = We multiply the features by two more weight sets. The first predicts the means μ of the latent distributions, the second their standard deviations σ. = 4. A random offset = Let us sample ε from a standard normal, mean 0 and variance 1, and multiply it by σ. This is a random step away from the mean, scaled by how uncertain each feature is. = 5. Mean plus offset = We add the offset back onto μ, and these become the decoder's inputs. Keeping the randomness out in ε is the reparameterization trick: it lets gradients flow straight through the sampling. = 6. Decoder, layer 1 = Let us multiply by weights and biases and apply ReLU again. Here -4 is crossed out. = 7. Decoder, layer 2 = We multiply once more. The output Y is the decoder's attempt to rebuild X from the sampled distribution. = 8. Gradient for the mean = Let us push μ toward 0. A lot of math, the SGVB estimator, collapses the KL gradient to simply μ itself. = 9. Gradient for the standard deviation = We want σ to approach 1. = 10. And its formula = That same math simplifies the gradient to σ minus 1/σ. = 11. Reconstruction gradient = We want the reconstruction Y to match the input X. Mean squared error simplifies its gradient to Y minus X. Takeaway: the two gradients you just calculated each sit at the heart of a modern method, so one VAE teaches you both. The KL divergence is the penalty RLHF like GRPO uses to keep a fine-tuned model from drifting off its base. The reconstruction loss, plain mean squared error, is exactly what trains a diffusion model to denoise. Draw one VAE by hand and you have quietly learned the core of both. 💾 Save this post!
0:31

Variational Autoencoder by hand ✍️ ~ 11 steps walkthrough below A VAE learns the structure of your data, the mean and variance of its hidden features, and then generates new data from that structure. A GAN only learns to fool a discriminator. It can make convincing fakes without ever knowing what the data is really made of. That is the difference, and it is the whole reason VAEs matter. In 2024 ICLR gave its first ever Test of Time Award to the VAE paper, "Auto-Encoding Variational Bayes" by Diederik Kingma and Max Welling, ten years on. How does it work? Goal: encode three inputs into a distribution, sample from it, decode it back, and read every loss gradient off the page. = 1. Given = Three training examples X1, X2, X3, copied to the bottom as their own targets. Reconstructing your own input is what puts the "auto", meaning self, in autoencoder. = 2. Encoder, layer 1 = Let us multiply the inputs by weights and biases, then apply ReLU, crossing out every negative. = 3. Mean and standard deviation = We multiply the features by two more weight sets. The first predicts the means μ of the latent distributions, the second their standard deviations σ. = 4. A random offset = Let us sample ε from a standard normal, mean 0 and variance 1, and multiply it by σ. This is a random step away from the mean, scaled by how uncertain each feature is. = 5. Mean plus offset = We add the offset back onto μ, and these become the decoder's inputs. Keeping the randomness out in ε is the reparameterization trick: it lets gradients flow straight through the sampling. = 6. Decoder, layer 1 = Let us multiply by weights and biases and apply ReLU again. Here -4 is crossed out. = 7. Decoder, layer 2 = We multiply once more. The output Y is the decoder's attempt to rebuild X from the sampled distribution. = 8. Gradient for the mean = Let us push μ toward 0. A lot of math, the SGVB estimator, collapses the KL gradient to simply μ itself. = 9. Gradient for the standard deviation = We want σ to approach 1. = 10. And its formula = That same math simplifies the gradient to σ minus 1/σ. = 11. Reconstruction gradient = We want the reconstruction Y to match the input X. Mean squared error simplifies its gradient to Y minus X. Takeaway: the two gradients you just calculated each sit at the heart of a modern method, so one VAE teaches you both. The KL divergence is the penalty RLHF like GRPO uses to keep a fine-tuned model from drifting off its base. The reconstruction loss, plain mean squared error, is exactly what trains a diffusion model to denoise. Draw one VAE by hand and you have quietly learned the core of both. 💾 Save this post!

Tom Yeh

17,011 次观看 • 2 个月前

Scale alone is not enough for AI data. Quality and complexity are equally critical. Excited to support all of these for LLM developers with Snorkel AI Data-as-a-Service, and to share our new leaderboard! — Our decade-plus of research and work in AI data has a simple point: scale alone is not enough. AI success is all about the quality, complexity, and distribution of data—in addition to volume. We’re excited to be powering leading LLM developers with Snorkel AI Expert Data-as-a-Service, our white glove service for custom, expert-level AI datasets—and to now preview some of what we’re building via our new Expert Data Leaderboard (🔗 in 🧵) + upcoming OSS dataset releases! Snorkel Expert Data-as-a-Service is built to meet the rapidly evolving data needs of the agentic AI world—where success is built on the quality, complexity, and distribution of datasets, in addition to size and scale. This kind of high-quality, frontier AI data can only come from a union of technology and human expertise. With Snorkel Expert Data-as-a-Service, we’re powering frontier LLM developers across agentic, expert knowledge, reasoning, coding, multi-modal, and other task types via the combination of these two key components: - (1) The Snorkel Expert Network: A global team of subject matter experts focused wholly on specialized knowledge–spanning thousands of topics in STEM/academic, vertical/professional, and consumer/lifestyle domains. - (2) Snorkel AI Data Development Platform: Our unique programmatic data curation and quality control platform, accelerating and improving expert authoring and review through principled techniques developed over the last decade of R&D. Now: we’re incredibly excited to showcase some of the power of Snorkel Expert Data-as-a-Service via the new Snorkel Leaderboard—putting frontier models to the test in complex, agentic, and reasoning settings inspired by real industry scenarios (not esoteric puzzles)! We’ll be releasing new leaderboards and accompanying expert-verified open source datasets (coming soon!) regularly. To start, we’re sharing three initial ones in preview: - SnorkelFinance: Q&A over financial documents requiring agentic tool-calling and reasoning - SnorkelUnderwrite: Agentic insurance tasks requiring industry-specific reasoning and tool use - SnorkelSequences: Mathematical tasks requiring compositional multi-step reasoning
1:11

Scale alone is not enough for AI data. Quality and complexity are equally critical. Excited to support all of these for LLM developers with Snorkel AI Data-as-a-Service, and to share our new leaderboard! — Our decade-plus of research and work in AI data has a simple point: scale alone is not enough. AI success is all about the quality, complexity, and distribution of data—in addition to volume. We’re excited to be powering leading LLM developers with Snorkel AI Expert Data-as-a-Service, our white glove service for custom, expert-level AI datasets—and to now preview some of what we’re building via our new Expert Data Leaderboard (🔗 in 🧵) + upcoming OSS dataset releases! Snorkel Expert Data-as-a-Service is built to meet the rapidly evolving data needs of the agentic AI world—where success is built on the quality, complexity, and distribution of datasets, in addition to size and scale. This kind of high-quality, frontier AI data can only come from a union of technology and human expertise. With Snorkel Expert Data-as-a-Service, we’re powering frontier LLM developers across agentic, expert knowledge, reasoning, coding, multi-modal, and other task types via the combination of these two key components: - (1) The Snorkel Expert Network: A global team of subject matter experts focused wholly on specialized knowledge–spanning thousands of topics in STEM/academic, vertical/professional, and consumer/lifestyle domains. - (2) Snorkel AI Data Development Platform: Our unique programmatic data curation and quality control platform, accelerating and improving expert authoring and review through principled techniques developed over the last decade of R&D. Now: we’re incredibly excited to showcase some of the power of Snorkel Expert Data-as-a-Service via the new Snorkel Leaderboard—putting frontier models to the test in complex, agentic, and reasoning settings inspired by real industry scenarios (not esoteric puzzles)! We’ll be releasing new leaderboards and accompanying expert-verified open source datasets (coming soon!) regularly. To start, we’re sharing three initial ones in preview: - SnorkelFinance: Q&A over financial documents requiring agentic tool-calling and reasoning - SnorkelUnderwrite: Agentic insurance tasks requiring industry-specific reasoning and tool use - SnorkelSequences: Mathematical tasks requiring compositional multi-step reasoning

Alex Ratner

495,851 次观看 • 1 年前

Day 9/90 of Inference Engineering I built a visualization tool for CUDA kernels and C code to view the buffers for each step through. Check out Yesterday, I was terramogged by a simple implementation of a basic matmul between three matrices in 1D vector form. There were essentially three 1D vectors, each vector representing a corresponding matrix (A, B, C). In the implementation, each matrix is just just a pointer to the first element of a 1D array. You are to compute the output matrix through A^T @ B and store it in C. This morning I built a visualizer to help me see and understand why the triple for-loops work the way they do. In the coming days, I’m confident I can make sense as well as backpropagation! As for it does not include a built-in compiler. It's just an interpreter; the visualization shows exactly what the code does, bugs included. The main thing I wanted to visualize is matmul with a triple for-loop between a sequential program to that of a parallel one. Expect bugs and other inconsistencies within I’ll polish this mini-project in the coming days. As for the rest of today, I’ll be trying to understand backpropagation in C and CUDA .
0:20

Day 9/90 of Inference Engineering I built a visualization tool for CUDA kernels and C code to view the buffers for each step through. Check out Yesterday, I was terramogged by a simple implementation of a basic matmul between three matrices in 1D vector form. There were essentially three 1D vectors, each vector representing a corresponding matrix (A, B, C). In the implementation, each matrix is just just a pointer to the first element of a 1D array. You are to compute the output matrix through A^T @ B and store it in C. This morning I built a visualizer to help me see and understand why the triple for-loops work the way they do. In the coming days, I’m confident I can make sense as well as backpropagation! As for it does not include a built-in compiler. It's just an interpreter; the visualization shows exactly what the code does, bugs included. The main thing I wanted to visualize is matmul with a triple for-loop between a sequential program to that of a parallel one. Expect bugs and other inconsistencies within I’ll polish this mini-project in the coming days. As for the rest of today, I’ll be trying to understand backpropagation in C and CUDA .

max fu

12,746 次观看 • 2 个月前

Why did Ian Whiffin agree to give expert testimony on Jen McCabe’s cellphone extraction, when the state refused to let him look at or even give him the full, original extraction OR its verification hash? A thread🧵 Full cellphone extractions, sometimes called forensic images, generate what’s called a “hash value”, which serves as a unique digital fingerprint necessary for ensuring the integrity of data. Any discrepancy between the hash values indicates tampering with or corruption of evidence, alerting forensic examiners to potential issues with the evidence. Hash verification is a fundamental principle and a rather ubiquitous practice in the world of digital forensics, where data validation and verification are key. It is the gold standard across the industry, and has also become so in the courtroom, whereby admissibility of digital evidence is determined by its relevance, authenticity and reliability. In court, the hash value can be used to demonstrate that the evidence has not been altered since its collection, and is a universal way for experts to authenticate and validate the reliability of data for the trial Court. But, an extraction that’s missing a hash value altogether is a huge red flag. 🚩 Who removed the hash value? And why? It’s necessary to the chain of custody, and as Gaurino and Tully would be well aware, it’s also an element of the data that an expert would require in order to verify and validate it. There’s no “good” reason for why someone would remove a hash value, and the extraction can’t be characterized as a forensic image as its origin is unknown. This was a deliberate step taken to hide something, which one could argue shows consciousness of guilt. If the data are true and accurate, why would you bring their integrity into question by removing the hash? However, if the data were altered or tampered with, and let’s say, hypothetically speaking, you wanted a digital forensic expert to provide testimony supportive of your “Google search” theory. . . In that hypothetical, you’d have to remove the hash value. Otherwise, the expert would immediately detect that the data were altered, as they would not be able to verify the hash against the original. #KarenReadTrial #JusticeForJohnOKeefe #FreeKarenRead #CantonCoverup #PoliceCorruption #KarenRead #Cellebrite #DFIR
0:23

Why did Ian Whiffin agree to give expert testimony on Jen McCabe’s cellphone extraction, when the state refused to let him look at or even give him the full, original extraction OR its verification hash? A thread🧵 Full cellphone extractions, sometimes called forensic images, generate what’s called a “hash value”, which serves as a unique digital fingerprint necessary for ensuring the integrity of data. Any discrepancy between the hash values indicates tampering with or corruption of evidence, alerting forensic examiners to potential issues with the evidence. Hash verification is a fundamental principle and a rather ubiquitous practice in the world of digital forensics, where data validation and verification are key. It is the gold standard across the industry, and has also become so in the courtroom, whereby admissibility of digital evidence is determined by its relevance, authenticity and reliability. In court, the hash value can be used to demonstrate that the evidence has not been altered since its collection, and is a universal way for experts to authenticate and validate the reliability of data for the trial Court. But, an extraction that’s missing a hash value altogether is a huge red flag. 🚩 Who removed the hash value? And why? It’s necessary to the chain of custody, and as Gaurino and Tully would be well aware, it’s also an element of the data that an expert would require in order to verify and validate it. There’s no “good” reason for why someone would remove a hash value, and the extraction can’t be characterized as a forensic image as its origin is unknown. This was a deliberate step taken to hide something, which one could argue shows consciousness of guilt. If the data are true and accurate, why would you bring their integrity into question by removing the hash? However, if the data were altered or tampered with, and let’s say, hypothetically speaking, you wanted a digital forensic expert to provide testimony supportive of your “Google search” theory. . . In that hypothetical, you’d have to remove the hash value. Otherwise, the expert would immediately detect that the data were altered, as they would not be able to verify the hash against the original. #KarenReadTrial #JusticeForJohnOKeefe #FreeKarenRead #CantonCoverup #PoliceCorruption #KarenRead #Cellebrite #DFIR

Olivia

270,376 次观看 • 2 年前

The Everywhere Edition is here: If you are on Shopify, you are everywhere your customers are shopping. This Edition features some key updates to Agentic Storefronts, Catalog and UCP because we want to be clear that if you want on Shopify you are everywhere by default. Every Edition is intentionally overwhelming. It’s a celebration of the progress we’ve made each season and serves as a complete list of features for those who want to pay attention. To make it easier to parse, I broke down some of my favorite features among the 200+ and explained some of the ambition behind them.
7:45

The Everywhere Edition is here: If you are on Shopify, you are everywhere your customers are shopping. This Edition features some key updates to Agentic Storefronts, Catalog and UCP because we want to be clear that if you want on Shopify you are everywhere by default. Every Edition is intentionally overwhelming. It’s a celebration of the progress we’ve made each season and serves as a complete list of features for those who want to pay attention. To make it easier to parse, I broke down some of my favorite features among the 200+ and explained some of the ambition behind them.

Vanessa Lee

14,422 次观看 • 3 个月前

Backpropagation by hand ✍️ ~ 11 steps walkthrough below Backpropagation is the algorithm that actually trains a neural network, and it is where most people stop following along. It is not calculus you cannot do. It is matrix multiplication, working backward, one layer at a time. So I drew and calculated one entirely by hand. Goal: push the loss gradient back through a 3-layer network and land on a new value for every weight and bias. = 1. Given = A 3-layer perceptron, an input X, predictions Ypred = [0.5, 0.5, 0], and the truth Ytarget = [0, 1, 0]. = 2. Backprop gradient cells = Let us draw empty cells for every gradient we are about to compute. The shape of the answer comes first. = 3. Layer 3 softmax = We get dL/dz3 straight from Ypred minus Ytarget = [0.5, -0.5, 0]. No chain rule needed, and that shortcut is the whole reason softmax and cross-entropy are paired. = 4. Layer 3 weights and biases = Let us multiply dL/dz3 by [a2 | 1]. One multiplication gives the gradient for W3 and b3 together. = 5. Layer 2 activations = We multiply dL/dz3 by W3 to get dL/da2. The gradient moves back across a layer the same way the signal moved forward. = 6. Layer 2 ReLU = Let us pass it through the gate: keep the gradient where the activation was positive, zero it everywhere else. = 7. Layer 2 weights and biases = We multiply dL/dz2 by [a1 | 1]. The same figure as step 4, one layer up. = 8. Layer 1 activations = Let us multiply dL/dz2 by W2. = 9. Layer 1 ReLU = We apply the same gate again, now on a1. = 10. Layer 1 weights and biases = Let us multiply dL/dz1 by [x | 1], and every weight in the network now has a gradient. = 11. Update = We subtract, and the network has learned. In practice a learning rate scales this step. The gradients: dL/dz3 = [0.5, -0.5, 0] dL/da1 = [1, -2, 2, -1] dL/dz1 = [0, -2, 2, -1] The takeaway: matrix multiplication is all you need. Just like the forward pass, backpropagation is matrix multiplications end to end. You can do every one by hand, slowly and imperfectly, which is exactly why a GPU's ability to do them fast mattered so much to deep learning. 💾 Save this post!
0:26

Backpropagation by hand ✍️ ~ 11 steps walkthrough below Backpropagation is the algorithm that actually trains a neural network, and it is where most people stop following along. It is not calculus you cannot do. It is matrix multiplication, working backward, one layer at a time. So I drew and calculated one entirely by hand. Goal: push the loss gradient back through a 3-layer network and land on a new value for every weight and bias. = 1. Given = A 3-layer perceptron, an input X, predictions Ypred = [0.5, 0.5, 0], and the truth Ytarget = [0, 1, 0]. = 2. Backprop gradient cells = Let us draw empty cells for every gradient we are about to compute. The shape of the answer comes first. = 3. Layer 3 softmax = We get dL/dz3 straight from Ypred minus Ytarget = [0.5, -0.5, 0]. No chain rule needed, and that shortcut is the whole reason softmax and cross-entropy are paired. = 4. Layer 3 weights and biases = Let us multiply dL/dz3 by [a2 | 1]. One multiplication gives the gradient for W3 and b3 together. = 5. Layer 2 activations = We multiply dL/dz3 by W3 to get dL/da2. The gradient moves back across a layer the same way the signal moved forward. = 6. Layer 2 ReLU = Let us pass it through the gate: keep the gradient where the activation was positive, zero it everywhere else. = 7. Layer 2 weights and biases = We multiply dL/dz2 by [a1 | 1]. The same figure as step 4, one layer up. = 8. Layer 1 activations = Let us multiply dL/dz2 by W2. = 9. Layer 1 ReLU = We apply the same gate again, now on a1. = 10. Layer 1 weights and biases = Let us multiply dL/dz1 by [x | 1], and every weight in the network now has a gradient. = 11. Update = We subtract, and the network has learned. In practice a learning rate scales this step. The gradients: dL/dz3 = [0.5, -0.5, 0] dL/da1 = [1, -2, 2, -1] dL/dz1 = [0, -2, 2, -1] The takeaway: matrix multiplication is all you need. Just like the forward pass, backpropagation is matrix multiplications end to end. You can do every one by hand, slowly and imperfectly, which is exactly why a GPU's ability to do them fast mattered so much to deep learning. 💾 Save this post!

Tom Yeh

962,229 次观看 • 2 个月前