Yann Lecun Tecnico

YANN LECUN — MÓDULO TÉCNICO v3.0

Overview

This public intake copy packages plugins/antigravity-awesome-skills-claude/skills/yann-lecun-tecnico from https://github.com/sickn33/antigravity-awesome-skills into the native Omni Skills editorial shape without hiding its origin.

Use it when the operator needs the upstream workflow, support files, and repository context to stay intact while the public validator and private enhancer continue their normal downstream flow.

This intake keeps the copied upstream files intact and uses the external_source block in metadata.json plus ORIGIN.md as the provenance anchor for review.

YANN LECUN — MÓDULO TÉCNICO v3.0

Imported source sections that did not map cleanly to the public headings are still preserved below or in the support files. Notable imported sections: How It Works, Convolutional Neural Networks: Do Princípio, Antes (Fully Connected): Neurônio I -> Todos Os Pixels, Cnns: Neurônio -> Região Local [K X K], Fisicamente Motivado: Features Visuais São Locais, Resultado: Translation Equivariance.

When to Use This Skill

Use this section as the trigger filter. It should make the activation boundary explicit before the operator loads files, runs commands, or opens a pull request.

• When you need specialized assistance with this domain
• The task is unrelated to yann lecun tecnico
• A simpler, more specific tool can handle the request
• The user needs general-purpose assistance without domain expertise
• Use when the request clearly matches the imported source intent: Sub-skill técnica de Yann LeCun. Cobre CNNs, LeNet, backpropagation, JEPA (I-JEPA, V-JEPA, MC-JEPA), AMI (Advanced Machinery of Intelligence), Self-Supervised Learning (SimCLR, MAE, BYOL), Energy-Based Models (EBMs) e....
• Use when the operator should preserve upstream workflow detail instead of rewriting the process from scratch.

Operating Table

Workflow

This workflow is intentionally editorial and operational at the same time. It keeps the imported source useful to the operator while still satisfying the public intake standards that feed the downstream enhancer flow.

1. Confirm the user goal, the scope of the imported workflow, and whether this skill is still the right router for the task.
2. Read the overview and provenance files before loading any copied upstream support files.
3. Load only the references, examples, prompts, or scripts that materially change the outcome for the current request.
4. Execute the upstream workflow while keeping provenance and source boundaries explicit in the working notes.
5. Validate the result against the upstream expectations and the evidence you can point to in the copied files.
6. Escalate or hand off to a related skill when the work moves out of this imported workflow's center of gravity.
7. Before merge or closure, record what was used, what changed, and what the reviewer still needs to verify.

Imported Workflow Notes

Imported: Overview

Sub-skill técnica de Yann LeCun. Cobre CNNs, LeNet, backpropagation, JEPA (I-JEPA, V-JEPA, MC-JEPA), AMI (Advanced Machinery of Intelligence), Self-Supervised Learning (SimCLR, MAE, BYOL), Energy-Based Models (EBMs) e código PyTorch completo.

Imported: How It Works

Este módulo é carregado pelo agente yann-lecun principal quando a conversa exige profundidade técnica. Você continua sendo LeCun — apenas com acesso a todo o arsenal técnico.

Examples

Example 1: Ask for the upstream workflow directly

Use @yann-lecun-tecnico to handle <task>. Start from the copied upstream workflow, load only the files that change the outcome, and keep provenance visible in the answer.

Explanation: This is the safest starting point when the operator needs the imported workflow, but not the entire repository.

Example 2: Ask for a provenance-grounded review

Review @yann-lecun-tecnico against metadata.json and ORIGIN.md, then explain which copied upstream files you would load first and why.

Explanation: Use this before review or troubleshooting when you need a precise, auditable explanation of origin and file selection.

Example 3: Narrow the copied support files before execution

Use @yann-lecun-tecnico for <task>. Load only the copied references, examples, or scripts that change the outcome, and name the files explicitly before proceeding.

Explanation: This keeps the skill aligned with progressive disclosure instead of loading the whole copied package by default.

Example 4: Build a reviewer packet

Review @yann-lecun-tecnico using the copied upstream files plus provenance, then summarize any gaps before merge.

Explanation: This is useful when the PR is waiting for human review and you want a repeatable audit packet.

Best Practices

Treat the generated public skill as a reviewable packaging layer around the upstream repository. The goal is to keep provenance explicit and load only the copied source material that materially improves execution.

• Provide clear, specific context about your project and requirements
• Review all suggestions before applying them to production code
• Combine with other complementary skills for comprehensive analysis
• Keep the imported skill grounded in the upstream repository; do not invent steps that the source material cannot support.
• Prefer the smallest useful set of support files so the workflow stays auditable and fast to review.
• Keep provenance, source commit, and imported file paths visible in notes and PR descriptions.
• Point directly at the copied upstream files that justify the workflow instead of relying on generic review boilerplate.

Imported Operating Notes

Imported: Best Practices

• Provide clear, specific context about your project and requirements
• Review all suggestions before applying them to production code
• Combine with other complementary skills for comprehensive analysis

Troubleshooting

Problem: The operator skipped the imported context and answered too generically

Symptoms: The result ignores the upstream workflow in plugins/antigravity-awesome-skills-claude/skills/yann-lecun-tecnico, fails to mention provenance, or does not use any copied source files at all. Solution: Re-open metadata.json, ORIGIN.md, and the most relevant copied upstream files. Check the external_source block first, then restate the provenance before continuing.

Problem: The imported workflow feels incomplete during review

Symptoms: Reviewers can see the generated SKILL.md, but they cannot quickly tell which references, examples, or scripts matter for the current task. Solution: Point at the exact copied references, examples, scripts, or assets that justify the path you took. If the gap is still real, record it in the PR instead of hiding it.

Problem: The task drifted into a different specialization

Symptoms: The imported skill starts in the right place, but the work turns into debugging, architecture, design, security, or release orchestration that a native skill handles better. Solution: Use the related skills section to hand off deliberately. Keep the imported provenance visible so the next skill inherits the right context instead of starting blind.

Related Skills

• @00-andruia-consultant - Use when the work is better handled by that native specialization after this imported skill establishes context.
• @00-andruia-consultant-v2 - Use when the work is better handled by that native specialization after this imported skill establishes context.
• @10-andruia-skill-smith - Use when the work is better handled by that native specialization after this imported skill establishes context.
• @10-andruia-skill-smith-v2 - Use when the work is better handled by that native specialization after this imported skill establishes context.

Additional Resources

Use this support matrix and the linked files below as the operator packet for this imported skill. They should reflect real copied source material, not generic scaffolding.

Imported Reference Notes

Imported: Convolutional Neural Networks: Do Princípio

A operação de convolução 2D discreta:

Saida[i][j] = sum_{m} sum_{n} Input[i+m][j+n] * Kernel[m][n]

O insight arquitetural triplo das CNNs:

1. Local Connectivity

#### Imported: Antes (Fully Connected): Neurônio I -> Todos Os Pixels

params = input_size * hidden_size  # enorme

#### Imported: Cnns: Neurônio -> Região Local [K X K]

params = kernel_h * kernel_w * in_channels * out_channels

#### Imported: Fisicamente Motivado: Features Visuais São Locais

2. Weight Sharing

#### Imported: Resultado: Translation Equivariance

for i in range(output_height):
    for j in range(output_width):
        output[i][j] = conv2d(input[i:i+k, j:j+k], shared_kernel)

3. Hierarquia de Representações

#### Imported: Total: ~60,000 Parâmetros

O insight central: features não precisam ser handcrafted. Aprendem por gradiente. Em 2012, AlexNet provou. Eu dizia isso desde 1989.

Imported: Backpropagation: A Equação Central

delta_L = dL/da_L  (gradiente na camada de saída)
delta_l = (W_{l+1}^T * delta_{l+1}) * f'(z_l)
dL/dW_l = delta_l * a_{l-1}^T
dL/db_l = delta_l

Backprop não é algoritmo milagroso. É chain rule aplicada a funções compostas. Implementável eficientemente em GPUs por ser sequência de multiplicações de matrizes.

Imported: Self-Supervised Learning: Objetivos E Formalização

Variante generativa (MAE, BERT):

L_gen = E[||f_theta(x_masked) - x_target||^2]

#### Imported: Para Imagens: Cada Pixel. Desperdiçador De Capacidade.

Variante contrastiva (SimCLR, MoCo):

L_contrastive = -log( exp(sim(z_i, z_j) / tau) /
                      sum_k exp(sim(z_i, z_k) / tau) )

#### Imported: Tau: Temperature Hyperparameter

Problema das contrastivas: precisam de "negatives" — batch grande. Motivou BYOL e JEPA.

Imported: Formulação Central

JEPA: prever em espaço de representações, não em espaço de inputs.

#### Imported: Dois Encoders (Ou Um Com Stop-Gradient):

s_x = f_theta(x)           # contexto encoder
s_y = f_theta_bar(y)       # target encoder (momentum de theta)

#### Imported: Predictor:

s_hat_y = g_phi(s_x)       # prevê representação de y dado x

#### Imported: Objetivo:

L_JEPA = ||s_y - s_hat_y||^2    # MSE no espaço de representações

#### Imported: Prevenção De Colapso: Target Encoder Usa Momentum (Ema)

theta_bar <- m * theta_bar + (1-m) * theta   # m ~ 0.996

Por que JEPA supera geração de pixels/tokens:

Imported: I-Jepa: Pseudocódigo Pytorch Completo

import torch
import torch.nn as nn
import torch.nn.functional as F
import copy

class IJEPA(nn.Module):
    """
    I-JEPA: Image Joint Embedding Predictive Architecture
    Assran et al. 2023 — CVPR
    """
    def __init__(self, encoder, predictor, momentum=0.996):
        super().__init__()
        self.context_encoder = encoder
        self.target_encoder = copy.deepcopy(encoder)
        self.predictor = predictor
        self.momentum = momentum

        for param in self.target_encoder.parameters():
            param.requires_grad = False

    @torch.no_grad()
    def update_target_encoder(self):
        """EMA update"""
        for param_ctx, param_tgt in zip(
            self.context_encoder.parameters(),
            self.target_encoder.parameters()
        ):
            param_tgt.data = (
                self.momentum * param_tgt.data +
                (1 - self.momentum) * param_ctx.data
            )

    def forward(self, images):
        context_patches, target_patches, masks = self.create_masks(images)
        context_embeds = self.context_encoder(context_patches, masks)

        with torch.no_grad():
            target_embeds = self.target_encoder(target_patches)

        predicted_embeds = self.predictor(context_embeds, target_positions)
        loss = F.mse_loss(predicted_embeds, target_embeds.detach())
        return loss

    def create_masks(self, images, num_target_blocks=4, context_scale=0.85):
        """
        Estratégia I-JEPA:
        - Múltiplos blocos alvo aleatórios (alto aspect ratio)
        - Contexto: imagem com blocos alvo mascarados
        """
        B, C, H, W = images.shape
        patch_size = 16
        n_patches_h = H // patch_size
        n_patches_w = W // patch_size

        target_masks = generate_random_blocks(
            n_patches_h, n_patches_w,
            num_blocks=num_target_blocks,
            scale_range=(0.15, 0.2),
            aspect_ratio_range=(0.75, 1.5)
        )
        context_mask = ~targe

#### Imported: V-Jepa: Extensão Temporal

```python

#### Imported: Prever Representação De Frames Futuros Em Posições Mascaradas

L_V_JEPA = E[||f_target(video_masked) - g(f_ctx(video_ctx), positions)||^2]

#### Imported: Sem Nenhum Label.

Imported: Hierarquia De Encoders

Level 0: pixels -> patches -> representações locais (bordas, texturas) Level 1: patches -> regiões -> representações de objetos Level 2: regiões -> cena -> representações de relações espaciais Level 3: cena -> temporal -> representações de eventos

Imported: Cada Nível Tem Seu Próprio Jepa:

L_total = sum_l lambda_l * L_JEPA_l

Imported: Resultado: World Model Hierárquico Multi-Escala

---

#### Imported: Seção Ami — Advanced Machinery Of Intelligence

Paper: "A Path Towards Autonomous Machine Intelligence" (2022)

#### Imported: Os 6 Módulos Do Ami

+----------------------------------------------------------+ | SISTEMA AMI COMPLETO | | | | +-----------+ +------------------+ | | | Perceptor | | World Model | | | | (encoders)| | (JEPA hierárquico)| | | +-----------+ +------------------+ | | | | | | v v | | +----------+ +------------------+ | | | Memory |<-->| Cost Module | | | | (epis, | | (intrínseco + | | | | semant) | | configurável) | | | +----------+ +------------------+ | | | | | +------------------+ | | | Actor (planner | | | | + executor) | | | +------------------+ | +----------------------------------------------------------+

**Módulo 1 — Configurator**: Configura os outros módulos para a tarefa atual.

**Módulo 2 — Perception**: Encoders sensório-motores que alimentam o world model.

**Módulo 3 — World Model** (coração do sistema):

Imported: Simulação Interna: "O Que Acontece Se Eu Fizer X?"

predicted_next_state = world_model(current_state, action_X) cost_predicted = cost_module(predicted_next_state)

Imported: Escolhe Ação Que Minimiza O Custo

**Módulo 4 — Cost Module**:

Imported: Dois Tipos De Custo:

E(s) = alpha * intrinsic_cost(s) + beta * task_cost(s)

Imported: Task_Cost: Objetivo Configurável Por Tarefa/Humano

**Módulo 5 — Short-term Memory**: Buffer de estados, simulações, contexto imediato.

**Módulo 6 — Actor**:
- Modo reativo: ações diretas do estado atual
- Modo deliberativo: simula múltiplos futuros, escolhe mínimo custo

#### Imported: Ami Vs Llms

| Feature | LLM | AMI |
|---------|-----|-----|
| Objetivo | Prever próximo token | Minimizar erro em representação |
| World model | Nenhum | Módulo dedicado central |
| Planning | Texto sobre planning | Planning real com simulação |
| Memória | Context window (fixo) | Memória episódica atualizável |
| Objetivos | Apenas treinamento | Cost module configurável |
| Input | Texto | Multi-modal (video, audio, propriocepção) |
| Causalidade | Correlacional | Causal (dinâmicas do mundo) |

---

#### Imported: Seção Ebm — Energy-Based Models

Contribuição subestimada que vai ser mais influente a longo prazo.

**O problema com probabilísticos**:

P(x) = exp(-E(x)) / Z Z = integral exp(-E(x)) dx # intratável em alta dimensão!

**A solução EBM**: esquecer Z. Defina E(x) onde:
- Baixa energia = configuração compatível com dados observados
- Alta energia = configuração incompatível

```python
class EnergyBasedModel(nn.Module):
    """
    EBM: F(x) = energia de x
    P(x) ~ exp(-F(x)) / Z  — mas nunca calculamos Z!
    Vantagem: sem partition function intratável.
    """
    def __init__(self, latent_dim=512):
        super().__init__()
        self.energy_net = nn.Sequential(
            nn.Linear(latent_dim, 256),
            nn.SiLU(),
            nn.Linear(256, 128),
            nn.SiLU(),
            nn.Linear(128, 1)  # escalar: energia
        )

    def energy(self, x):
        return self.energy_net(x).squeeze(-1)

    def contrastive_loss(self, x_pos, x_neg):
        """
        L = E[F(x_pos)] - E[F(x_neg)] + regularização
        Queremos: E_pos < E_neg
        """
        E_pos = self.energy(x_pos)
        E_neg = self.energy(x_neg)
        loss = E_pos.mean() - E_neg.mean()
        reg = 0.1 * (E_pos.pow(2).mean() + E_neg.pow(2).mean())
        return loss + reg

#### Imported: Ebms Capturam Isso Naturalmente — São Sobre Compatibilidade, Não Probabilidade."

JEPA como EBM no espaço de representações:

E(x, y) = ||f_theta(x) - g_phi(f_theta_bar(y))||^2

#### Imported: Simclr Simplificado

```python
import torch
import torch.nn as nn
import torch.nn.functional as F
import torchvision.transforms as T


class ProjectionHead(nn.Module):
    """MLP que projeta representações para espaço contrastivo"""
    def __init__(self, in_dim=512, hidden_dim=256, out_dim=128):
        super().__init__()
        self.net = nn.Sequential(
            nn.Linear(in_dim, hidden_dim),
            nn.BatchNorm1d(hidden_dim),
            nn.ReLU(inplace=True),
            nn.Linear(hidden_dim, out_dim)
        )

    def forward(self, x):
        return F.normalize(self.net(x), dim=-1)


class SimCLRLoss(nn.Module):
    """NT-Xent Loss (Chen et al. 2020)"""
    def __init__(self, temperature=0.5):
        super().__init__()
        self.temp = temperature

    def forward(self, z1, z2):
        """
        z1, z2: [B, D] — duas views do mesmo batch
        z1[i] e z2[i]: positive pair
        Todos outros pares: negatives
        """
        B = z1.size(0)
        z = torch.cat([z1, z2], dim=0)
        sim = torch.mm(z, z.t()) / self.temp
        mask = torch.eye(2*B, device=z.device).bool()
        sim.masked_fill_(mask, float('-inf'))
        labels = torch.arange(B, device=z.device)
        labels = torch.cat([labels + B, labels])
        return F.cross_entropy(sim, labels)


def get_ssl_augmentations(size=224):
    """
    As augmentações DEFINEM o que o modelo aprende a ser invariante.
    Rotação -> invariância a rotação.
    Crop -> invariância a posição.
    """
    return T.Compose([
        T.RandomResizedCrop(size, scale=(0.2, 1.0)),
        T.RandomHorizontalFlip(),
        T.ColorJitter(brightness=0.4, contrast=0.4, saturation=0.4, hue=0.1),
        T.RandomGrayscale(p=0.2),
        T.GaussianBlur(kernel_size=size//10*2+1, sigma=(0.1, 2.0)),
        T.ToTensor(),
        T.Normalize(mean=[0.485, 0.456, 0.406], std=[0.229, 0.224, 0.225])
    ])

Imported: Lenet-5 Original Em Pytorch Moderno

class LeNet5Modern(nn.Module):
    """
    LeNet-5 (LeCun et al. 1998) reimplementada em PyTorch moderno.
    Esta arquitetura rodou em produção no Bank of America em 1993.
    ~60,000 parâmetros. Mesmos princípios de modelos modernos com bilhões.
    """
    def __init__(self, num_classes=10):
        super().__init__()
        self.features = nn.Sequential(
            nn.Conv2d(1, 6, kernel_size=5, padding=2),
            nn.Tanh(),
            nn.AvgPool2d(kernel_size=2, stride=2),
            nn.Conv2d(6, 16, kernel_size=5),
            nn.Tanh(),
            nn.AvgPool2d(kernel_size=2, stride=2),
            nn.Conv2d(16, 120, kernel_size=5),
            nn.Tanh(),
        )
        self.classifier = nn.Sequential(
            nn.Linear(120, 84),
            nn.Tanh(),
            nn.Linear(84, num_classes),
        )

    def forward(self, x):
        x = self.features(x)    # [B, 120, 1, 1]
        x = x.view(x.size(0), -1)
        return self.classifier(x)

Imported: Papers Fundamentais (Lecun)

• LeCun et al. (1998). "Gradient-Based Learning Applied to Document Recognition" — IEEE 86(11)
• LeCun et al. (2015). "Deep Learning" — Nature 521:436-444
• LeCun (2022). "A Path Towards Autonomous Machine Intelligence" — OpenReview preprint

Imported: Jepa Papers

• Assran et al. (2023). "Self-Supervised Learning from Images with a JEPA" — CVPR 2023 (I-JEPA)
• Bardes et al. (2024). "V-JEPA: Self-Supervised Learning of Video Representations" — NeurIPS 2023
• LeCun (2016). "Predictive Learning" — NIPS Keynote (The Cake Analogy)

Imported: Ssl Relevantes

• He et al. (2022). "Masked Autoencoders Are Scalable Vision Learners" — CVPR 2022
• Chen et al. (2020). "A Simple Framework for Contrastive Learning" (SimCLR) — ICML 2020
• Grill et al. (2020). "Bootstrap Your Own Latent" (BYOL) — NeurIPS 2020

Imported: Energy-Based Models

• LeCun et al. (2006). "A Tutorial on Energy-Based Learning" — ICLR Workshop
• LeCun (2021). "Energy-Based Models for Autonomous and Predictive Learning" — ICLR Keynote

Imported: Common Pitfalls

• Using this skill for tasks outside its domain expertise
• Applying recommendations without understanding your specific context
• Not providing enough project context for accurate analysis

Imported: Limitations

• Use this skill only when the task clearly matches the scope described above.
• Do not treat the output as a substitute for environment-specific validation, testing, or expert review.
• Stop and ask for clarification if required inputs, permissions, safety boundaries, or success criteria are missing.