Technical blog on Artificial Intelligence, Embodied Intelligence, Robotics, Computer Vision, 3D Reconstruction, LLMs, and AGI.https://ailinkdeeptech.com/https://ailinkdeeptech.com/research/ppo_algo/https://ailinkdeeptech.com/research/ppo_algo/A PyTorch implementation of Proximal Policy Optimization (PPO) with a clipped surrogate objective, shared actor-critic network, Gaussian policy, and value/entropy losses.Wed, 14 Oct 2026 00:00:00 GMThttps://ailinkdeeptech.com/research/sac_algo/https://ailinkdeeptech.com/research/sac_algo/A PyTorch implementation of Soft Actor-Critic (SAC) with a stochastic squashed Gaussian policy, twin Q-networks with target networks, automatic temperature tuning, and a replay buffer for continuous control.Thu, 24 Sep 2026 00:00:00 GMThttps://ailinkdeeptech.com/research/nerf_algo/https://ailinkdeeptech.com/research/nerf_algo/A PyTorch implementation of Neural Radiance Fields (NeRF): positional encoding for positions and view directions, MLP with skip connections, density/color prediction, stratified sampling, and volumetric rendering.Tue, 01 Sep 2026 00:00:00 GMThttps://ailinkdeeptech.com/research/rope_algo/https://ailinkdeeptech.com/research/rope_algo/A PyTorch implementation of Rotary Position Embedding (RoPE): inverse frequency buffer, precomputed cos/sin cache, query/key rotation, and shape verification.Sun, 02 Aug 2026 00:00:00 GMThttps://ailinkdeeptech.com/articles/multi-agent-embodied-systems-2026/https://ailinkdeeptech.com/articles/multi-agent-embodied-systems-2026/Master multi-agent embodied systems: CTDE reinforcement learning, shared 3DGS maps, dynamic role allocation, and heterogeneous robot team deployment.Fri, 17 Jul 2026 10:00:00 GMThttps://ailinkdeeptech.com/articles/world-models-embodied-ai-tutorial-2026/https://ailinkdeeptech.com/articles/world-models-embodied-ai-tutorial-2026/Master World Models for Embodied AI: Recurrent State Space Models (RSSM), latent dynamics, model-based MPC planning, and VLA policy integration in PyTorch.Fri, 17 Jul 2026 10:00:00 GMThttps://ailinkdeeptech.com/research/stablediffusion_algo/https://ailinkdeeptech.com/research/stablediffusion_algo/A minimal PyTorch implementation of a Denoising Diffusion Probabilistic Model (DDPM): sinusoidal time embeddings, U-Net with skip connections, forward noising, reverse denoising, and ancestral sampling.Wed, 24 Jun 2026 00:00:00 GMThttps://ailinkdeeptech.com/research/nerf/https://ailinkdeeptech.com/research/nerf/Master NeRF (Neural Radiance Fields): 5D radiance fields, differentiable volume rendering, Fourier positional encoding, hierarchical sampling, and PyTorch.Mon, 22 Jun 2026 12:36:00 GMThttps://ailinkdeeptech.com/cookbook/qwen3_vl_8b_vision_grpo/https://ailinkdeeptech.com/cookbook/qwen3_vl_8b_vision_grpo/Train Qwen3-VL-8B Vision with GRPO: multimodal policy optimization, two-tier visual reward functions, and 16-bit LoRA under 15GB VRAM.Thu, 18 Jun 2026 12:00:00 GMThttps://ailinkdeeptech.com/cookbook/qwen3_5_4b_vision/https://ailinkdeeptech.com/cookbook/qwen3_5_4b_vision/Fine-tune Qwen3.5-4B Vision using Unsloth: 16-bit LoRA optimization, multi-modal patch projection, and Q4_K_M GGUF edge deployment on consumer GPUs.Thu, 11 Jun 2026 11:37:00 GMThttps://ailinkdeeptech.com/articles/vision-language-action-vla-model-robotic-manipulation-tutorial-2026/https://ailinkdeeptech.com/articles/vision-language-action-vla-model-robotic-manipulation-tutorial-2026/Master Vision-Language-Action (VLA) models: continuous Flow Matching action heads, cross-embodiment datasets, Action Chunking, and Isaac Lab sim-to-real.Thu, 11 Jun 2026 10:00:00 GMThttps://ailinkdeeptech.com/research/transformer_algo/https://ailinkdeeptech.com/research/transformer_algo/A compact PyTorch implementation of a Transformer encoder with multi-head self-attention, sinusoidal positional encoding, feed-forward layers, residual connections, and padding masks.Thu, 04 Jun 2026 00:00:00 GMThttps://ailinkdeeptech.com/cookbook/flux2_klein_lora/https://ailinkdeeptech.com/cookbook/flux2_klein_lora/Train FLUX.2 Klein LoRA on 8GB GPUs using Kohya_ss: compact 4B MMDiT flow matching, CLIP-L conditioning, latent caching, and ComfyUI deployment.Wed, 03 Jun 2026 06:15:00 GMThttps://ailinkdeeptech.com/cookbook/qwen3_5_4b_vision_grpo/https://ailinkdeeptech.com/cookbook/qwen3_5_4b_vision_grpo/Train Qwen3.5-4B Vision with GRPO: multimodal policy optimization, two-tier visual reward functions, and 16-bit LoRA under 15GB VRAM.Mon, 01 Jun 2026 12:00:00 GMThttps://ailinkdeeptech.com/cookbook/gpt-oss-fine-tuning/https://ailinkdeeptech.com/cookbook/gpt-oss-fine-tuning/Fine-tune OpenAI gpt-oss-20B MoE using Unsloth on 16GB GPUs: 4-bit QLoRA, Harmony chat templates, channel separation, and SFTTrainer pipeline.Sat, 30 May 2026 07:20:00 GMThttps://ailinkdeeptech.com/cookbook/qwen3_vl_8b_vision/https://ailinkdeeptech.com/cookbook/qwen3_vl_8b_vision/Fine-tune Qwen3-VL-8B Vision on LaTeX OCR tasks with Unsloth: 4-bit NF4 QLoRA, multimodal token collation, and GGUF export under 8.5GB VRAM.Wed, 27 May 2026 12:03:00 GMThttps://ailinkdeeptech.com/articles/video-generation-foundation-models-2026/https://ailinkdeeptech.com/articles/video-generation-foundation-models-2026/Master Video Generation Foundation Models: 3D Diffusion Transformers, continuous Flow Matching, 3D VAE compression, and open-source PyTorch pipelines.Thu, 21 May 2026 10:00:00 GMThttps://ailinkdeeptech.com/cookbook/qwen3_5_0_8b_vision/https://ailinkdeeptech.com/cookbook/qwen3_5_0_8b_vision/Fine-tune Qwen3.5-0.8B Vision on VQA tasks using Unsloth: BF16 LoRA optimization, multi-modal token alignment, and GGUF edge deployment under 3GB VRAM.Tue, 19 May 2026 11:13:00 GMThttps://ailinkdeeptech.com/cookbook/qwen3_5_2b_vision/https://ailinkdeeptech.com/cookbook/qwen3_5_2b_vision/Fine-tune Qwen3.5-2B Vision on document VQA using Unsloth: 16-bit LoRA optimization, multi-modal OCR token alignment, and GGUF export under 5GB VRAM.Mon, 18 May 2026 11:30:00 GMThttps://ailinkdeeptech.com/articles/openclaw-self-hosted-personal-ai-agent-complete-2026-tutorial/https://ailinkdeeptech.com/articles/openclaw-self-hosted-personal-ai-agent-complete-2026-tutorial/Master OpenClaw: self-hosted gateway routing, Docker sandbox isolation, SKILL.md progressive disclosure, MCP servers, and multi-channel messaging deployment.Sat, 16 May 2026 10:00:00 GMThttps://ailinkdeeptech.com/cookbook/qwen3_4b_thinking/https://ailinkdeeptech.com/cookbook/qwen3_4b_thinking/Fine-tune Qwen3-4B-Thinking on DeepSeek-R1 CoT traces with Unsloth: OpenMathReasoning alignment, thinking-mode toggles, and GGUF multi-quant export.Fri, 15 May 2026 13:20:00 GMThttps://ailinkdeeptech.com/cookbook/qwen3_14_reasoning-conversational/https://ailinkdeeptech.com/cookbook/qwen3_14_reasoning-conversational/Fine-tune Qwen3-14B dual-mode reasoning and chat: 75/25 mixed-dataset distillation, 4-bit QLoRA, and GGUF multi-quant export on 12GB VRAM.Fri, 08 May 2026 12:00:00 GMThttps://ailinkdeeptech.com/cookbook/gpt-oss-20b-grpo/https://ailinkdeeptech.com/cookbook/gpt-oss-20b-grpo/Train gpt-oss-20B MoE using GRPO: AST sandboxing, anti-reward hacking, 4-bit QLoRA, and execution benchmark reward engineering.Sun, 03 May 2026 12:49:00 GMThttps://ailinkdeeptech.com/cookbook/flux1_dev_lora/https://ailinkdeeptech.com/cookbook/flux1_dev_lora/Master FLUX.1 Dev LoRA fine-tuning with Kohya_ss: MMDiT joint attention, Flow Matching velocity loss, text encoder freezing, and ComfyUI deployment.Fri, 01 May 2026 06:14:00 GMThttps://ailinkdeeptech.com/articles/trellis-structured-3d-latents-text-to-3d-microsoft-2026/https://ailinkdeeptech.com/articles/trellis-structured-3d-latents-text-to-3d-microsoft-2026/Master Microsoft TRELLIS: Structured 3D Latents (SLAT), Rectified Flow Transformers, and unified decoding into 3D Gaussian Splats, NeRFs, and meshes.Mon, 27 Apr 2026 10:00:00 GMThttps://ailinkdeeptech.com/cookbook/sdxl_lora/https://ailinkdeeptech.com/cookbook/sdxl_lora/Train high-fidelity SDXL LoRAs with Kohya_ss: latent diffusion objective, cross-attention projection tuning, multi-aspect bucketing, and ComfyUI deployment.Wed, 22 Apr 2026 06:13:00 GMThttps://ailinkdeeptech.com/cookbook/qwen3_4b_grpo/https://ailinkdeeptech.com/cookbook/qwen3_4b_grpo/Train Qwen3-4B with GRPO on 16GB GPUs: two-stage format pre-tuning, 4-tier composite reward shaping, rsLoRA, and vLLM acceleration.Tue, 21 Apr 2026 12:52:00 GMThttps://ailinkdeeptech.com/cookbook/qwen3_8b_fp8_grpo/https://ailinkdeeptech.com/cookbook/qwen3_8b_fp8_grpo/Train Qwen3-8B in native FP8 precision using GRPO reinforcement learning: memory-efficient policy updates, multi-tier reward functions, and 20GB VRAM execution.Tue, 21 Apr 2026 12:42:00 GMThttps://ailinkdeeptech.com/articles/hunyuan3d-2-tencent-high-resolution-text-to-3d-image-to-3d-2026/https://ailinkdeeptech.com/articles/hunyuan3d-2-tencent-high-resolution-text-to-3d-image-to-3d-2026/Master Tencent's Hunyuan3D-2: explore the decoupled Hunyuan3D-DiT shape generator, Hunyuan3D-Paint 2K PBR texture synthesis, and Blender pipeline.Fri, 17 Apr 2026 10:00:00 GMThttps://ailinkdeeptech.com/cookbook/qwen3_4b_instruct/https://ailinkdeeptech.com/cookbook/qwen3_4b_instruct/Fine-tune Qwen3-4B-Instruct with Unsloth: response-only loss masking, 4-bit QLoRA, memory profiling under 8GB VRAM, and GGUF multi-quant export.Wed, 15 Apr 2026 12:00:00 GMThttps://ailinkdeeptech.com/cookbook/qwen_image_lora/https://ailinkdeeptech.com/cookbook/qwen_image_lora/Train 20B Qwen-Image LoRA on 24GB GPUs using AI-Toolkit: uint3 quantization, Accuracy Recovery Adapters (ARA), Flow Matching, and ComfyUI deployment.Mon, 13 Apr 2026 07:00:00 GMThttps://ailinkdeeptech.com/articles/multimodal-agents-text-vision-action-embodied-intelligence-2026/https://ailinkdeeptech.com/articles/multimodal-agents-text-vision-action-embodied-intelligence-2026/Master multimodal embodied agents: Vision-Language-Action (VLA) backbones, flow-matching continuous control, hierarchical memory, and Isaac Lab sim-to-real.Sun, 12 Apr 2026 10:00:00 GMThttps://ailinkdeeptech.com/cookbook/qwen2_5_coder_1_5b_tool_calling/https://ailinkdeeptech.com/cookbook/qwen2_5_coder_1_5b_tool_calling/Fine-tune Qwen2.5-Coder-1.5B for tool calling using Unsloth: Hermes JSON schema formatting, 4-bit QLoRA, multi-tool dispatch, and edge deployment.Fri, 10 Apr 2026 13:07:00 GMThttps://ailinkdeeptech.com/cookbook/character_consistent_character_lora/https://ailinkdeeptech.com/cookbook/character_consistent_character_lora/Master LTX-2 Character-Consistent Video LoRA training: in-context conditioning (IC-LoRA), paired dataset curation, YAML configs, and ComfyUI deployment.Fri, 03 Apr 2026 07:50:00 GMThttps://ailinkdeeptech.com/cookbook/flux_2_dev_lora/https://ailinkdeeptech.com/cookbook/flux_2_dev_lora/Master FLUX.2-dev LoRA fine-tuning with AI-Toolkit: MMDiT flow matching, dataset curation, Prodigy adaptive optimization, and ComfyUI deployment.Fri, 03 Apr 2026 07:00:00 GMThttps://ailinkdeeptech.com/cookbook/qwen_image_edit_2511_lora/https://ailinkdeeptech.com/cookbook/qwen_image_edit_2511_lora/Train Qwen-Image-Edit-2511 LoRA using AI-Toolkit: paired MMDiT datasets, qfloat8 quantization, Diff Output Preservation, and ComfyUI deployment.Fri, 03 Apr 2026 07:00:00 GMThttps://ailinkdeeptech.com/cookbook/faster-moe/https://ailinkdeeptech.com/cookbook/faster-moe/Accelerate MoE LLM fine-tuning with Unsloth Faster MoE: torch._grouped_mm, Triton fused kernels, Split LoRA memory optimization, and benchmarks.Mon, 30 Mar 2026 10:38:00 GMThttps://ailinkdeeptech.com/articles/hunyuanvideo-tencent-text-to-video-generation-2026/https://ailinkdeeptech.com/articles/hunyuanvideo-tencent-text-to-video-generation-2026/Master Tencent's HunyuanVideo: 13B parameter flow-matching DiT, Causal 3D VAE compression, MLLM text conditioning, and PyTorch deployment.Tue, 17 Mar 2026 10:00:00 GMThttps://ailinkdeeptech.com/articles/real-time-3d-gaussian-splatting-tutorial-2026-5/https://ailinkdeeptech.com/articles/real-time-3d-gaussian-splatting-tutorial-2026-5/Master 3D Gaussian Splatting from scratch: anisotropic covariance decomposition, spherical harmonics, tile-based CUDA rasterization, and PyTorch training.Mon, 16 Mar 2026 10:00:00 GMThttps://ailinkdeeptech.com/articles/text-to-3d-gaussian-splatting-diffusion-2026/https://ailinkdeeptech.com/articles/text-to-3d-gaussian-splatting-diffusion-2026/Master Text-to-3D with Gaussian Splatting: Score Distillation Sampling (SDS), diffusion guidance, differentiable gsplat rendering, and PyTorch pipelines.Tue, 10 Mar 2026 10:00:00 GMThttps://ailinkdeeptech.com/cookbook/advanced_llama3_2_3b_grpo_lora/https://ailinkdeeptech.com/cookbook/advanced_llama3_2_3b_grpo_lora/Train Llama 3.2 (3B) with GRPO and LoRA: multi-reward optimization, SFT format warm-up, cosine similarity scoring, and PyTorch deployment.Sun, 01 Mar 2026 13:02:00 GMThttps://ailinkdeeptech.com/articles/hermes-agent-nous-research-self-improving-ai-agent-2026-tutorial/https://ailinkdeeptech.com/articles/hermes-agent-nous-research-self-improving-ai-agent-2026-tutorial/Master Hermes Agent by Nous Research: explore the closed learning loop, hierarchical memory, progressive skill discovery, and multi-backend MCP integration.Thu, 19 Feb 2026 10:00:00 GMThttps://ailinkdeeptech.com/articles/gs-slam-vla-robotic-manipulation-tutorial-2026/https://ailinkdeeptech.com/articles/gs-slam-vla-robotic-manipulation-tutorial-2026/Deploy GS-SLAM with VLA models for robotic manipulation: 3D Gaussian spatial grounding, cross-attention feature fusion, and closed-loop control.Wed, 18 Feb 2026 10:00:00 GMThttps://ailinkdeeptech.com/articles/scaffold-gs-structured-3d-gaussians-2026/https://ailinkdeeptech.com/articles/scaffold-gs-structured-3d-gaussians-2026/Master Scaffold-GS: sparse voxel anchor grids, shared neural MLP predictors, view-dependent LOD rendering, and 10x-100x Gaussian splat compression.Wed, 11 Feb 2026 14:00:00 GMThttps://ailinkdeeptech.com/cookbook/deepspeed_megatron/https://ailinkdeeptech.com/cookbook/deepspeed_megatron/Master distributed LLM pretraining (13B–70B+) with DeepSpeed and Megatron-LM: 3D Parallelism (TP/PP/DP), ZeRO-3, FP8 TransformerEngine, and multi-node NCCL.Tue, 03 Feb 2026 08:10:00 GMThttps://ailinkdeeptech.com/cookbook/llama_fp8_grpo/https://ailinkdeeptech.com/cookbook/llama_fp8_grpo/Train Llama 3.2 1B using hardware FP8 quantization and GRPO: 4-layer reward shaping, vLLM weight sharing, and 60% VRAM reduction.Sun, 01 Feb 2026 13:30:00 GMThttps://ailinkdeeptech.com/cookbook/llama3_1_8b_grpo/https://ailinkdeeptech.com/cookbook/llama3_1_8b_grpo/Train Llama 3.1 8B with GRPO using Unsloth on a single 16GB GPU: 5-tier composite reward functions, 4-bit QLoRA, XML reasoning, and vLLM acceleration.Sun, 11 Jan 2026 12:30:00 GMThttps://ailinkdeeptech.com/articles/gaussian-splatting-slam-gs-slam-real-time-dense-mapping-2026/https://ailinkdeeptech.com/articles/gaussian-splatting-slam-gs-slam-real-time-dense-mapping-2026/Master Gaussian Splatting SLAM (GS-SLAM): real-time pose tracking, incremental Gaussian mapping, loop closure, and PyTorch deployment on RGB-D streams.Sun, 11 Jan 2026 10:00:00 GMThttps://ailinkdeeptech.com/cookbook/deepseek_r1_0528_qwen3_8b_grpo/https://ailinkdeeptech.com/cookbook/deepseek_r1_0528_qwen3_8b_grpo/Train DeepSeek-R1 distilled Qwen3 (8B) using GRPO: multi-signal reward functions, group advantage normalization, and single-GPU fine-tuning.Thu, 01 Jan 2026 13:00:00 GMThttps://ailinkdeeptech.com/articles/4d-gaussian-splatting-dynamic-scenes-tutorial-2026/https://ailinkdeeptech.com/articles/4d-gaussian-splatting-dynamic-scenes-tutorial-2026/Master 4D Gaussian Splatting: canonical space formulation, HexPlane vs. MLP deformation fields, real-time PyTorch training, and robotics digital twins.Fri, 19 Dec 2025 09:00:00 GMThttps://ailinkdeeptech.com/articles/3d-gaussian-splatting-embodied-ai-robotics-2026/https://ailinkdeeptech.com/articles/3d-gaussian-splatting-embodied-ai-robotics-2026/Deploy 3D Gaussian Splatting in robotics: real-time scene reconstruction, Mahalanobis collision checking, 6-DoF grasp planning, and sim-to-real transfer.Wed, 17 Dec 2025 11:30:00 GMThttps://ailinkdeeptech.com/research/ltx-2/https://ailinkdeeptech.com/research/ltx-2/Master LTX-2: asymmetric dual-stream video/audio DiT, 1D/3D RoPE alignment, cross-modality adaLN, multimodal guidance (CFG/STG), and PyTorch pipelines.Sat, 13 Dec 2025 00:00:00 GMThttps://ailinkdeeptech.com/research/dpo_algo/https://ailinkdeeptech.com/research/dpo_algo/A PyTorch implementation of Direct Preference Optimization (DPO): a small LSTM language model, a preference dataset, and a DPO trainer that optimizes a policy against a frozen reference via the sigmoid logistic DPO loss on chosen/rejected pairs.Mon, 01 Dec 2025 00:00:00 GMThttps://ailinkdeeptech.com/research/grpo_algo/https://ailinkdeeptech.com/research/grpo_algo/A PyTorch implementation of Group Relative Policy Optimization (GRPO): a Group-relative RL approach that partitions sorted trajectories into groups, weights group-relative advantages, and applies clipped surrogate updates with PPO-style ratios.Mon, 24 Nov 2025 00:00:00 GMThttps://ailinkdeeptech.com/research/ddpg_algo/https://ailinkdeeptech.com/research/ddpg_algo/A PyTorch implementation of Deep Deterministic Policy Gradient (DDPG): an Actor that outputs tanh-bounded deterministic actions, a Q-value Critic over state-action pairs, a replay buffer, target networks with soft updates, and a smoke test that exercises action selection and a short training loop.Sat, 22 Nov 2025 00:00:00 GMThttps://ailinkdeeptech.com/research/llava_algo/https://ailinkdeeptech.com/research/llava_algo/A PyTorch implementation of LLaVA: a ResNet-50 vision encoder, a Transformer-based language model, a custom cross-attention fusion layer, and an enhanced tokenizer for image-grounded text generation.Sun, 09 Nov 2025 00:00:00 GMThttps://ailinkdeeptech.com/research/diffusion_transformer/https://ailinkdeeptech.com/research/diffusion_transformer/Master Diffusion Transformers (DiT): latent patchification, adaLN-Zero conditioning, multi-modal MMDiT joint attention, scaling laws, and PyTorch pipelines.Wed, 05 Nov 2025 00:00:00 GMThttps://ailinkdeeptech.com/research/diffusiontransformer_algo/https://ailinkdeeptech.com/research/diffusiontransformer_algo/A PyTorch implementation of the Diffusion Transformer (DiT): a patch-embedded Vision Transformer with sinusoidal timestep conditioning, multi-head self-attention, MLP blocks, and a transposed-conv head that predicts the image-space noise.Tue, 04 Nov 2025 00:00:00 GMThttps://ailinkdeeptech.com/research/dqn_algo/https://ailinkdeeptech.com/research/dqn_algo/A PyTorch implementation of Deep Q-Network (DQN): a Q-value MLP, a replay buffer, an ε-greedy agent with target network, Bellman loss update, and a CartPole-v1 training/test loop using Gymnasium.Sat, 01 Nov 2025 00:00:00 GMThttps://ailinkdeeptech.com/research/lora_algo/https://ailinkdeeptech.com/research/lora_algo/A PyTorch implementation of Low-Rank Adaptation (LoRA): low-rank A/B matrices wrapped around a frozen linear layer, scaled by alpha/rank, with shape, freezing, and rank-property tests.Mon, 27 Oct 2025 00:00:00 GMThttps://ailinkdeeptech.com/research/diffusion_algo/https://ailinkdeeptech.com/research/diffusion_algo/A PyTorch implementation of a Denoising Diffusion Probabilistic Model (DDPM): a linear-beta forward diffusion, a reverse sampling loop, a time-conditioned UNet denoising network, and a training-free smoke test that forward-diffuses, predicts noise, and samples images.Mon, 20 Oct 2025 00:00:00 GMThttps://ailinkdeeptech.com/research/flashattention_algo/https://ailinkdeeptech.com/research/flashattention_algo/A PyTorch implementation of FlashAttention-style multi-head attention: a softmax-scaled dot-product attention block, a tiled/blocked variant that processes sequence chunks for memory efficiency, and basic shape tests.Sun, 19 Oct 2025 00:00:00 GMThttps://ailinkdeeptech.com/research/moe_algo/https://ailinkdeeptech.com/research/moe_algo/A PyTorch implementation of the Mixture of Experts (MoE) layer with a top-k gating network, multiple expert MLPs, and weighted expert outputs for sparse expert routing.Fri, 17 Oct 2025 00:00:00 GMThttps://ailinkdeeptech.com/research/controlnet_algo/https://ailinkdeeptech.com/research/controlnet_algo/A minimal PyTorch ControlNet implementation: a control encoder with conv blocks, residual skip connections, channel attention, and a zero-initialized projection that adds residual control features into a backbone so that the model behaves as identity at init.Thu, 09 Oct 2025 00:00:00 GMThttps://ailinkdeeptech.com/research/ddim_algo/https://ailinkdeeptech.com/research/ddim_algo/A PyTorch implementation of a Denoising Diffusion Implicit Model (DDIM) sampler: a UNet denoiser with GELU time conditioning, an eta-controlled stochastic-to-deterministic reverse process on a sub-sequence of timesteps, and a smoke test that verifies noise addition, model output, sampling shapes, and deterministic reproducibility.Wed, 08 Oct 2025 00:00:00 GMThttps://ailinkdeeptech.com/research/ddpm_algo/https://ailinkdeeptech.com/research/ddpm_algo/A PyTorch implementation of a Denoising Diffusion Probabilistic Model (DDPM): a UNet denoiser with sinusoidal-free linear time conditioning, a linear-beta forward diffusion, an iterative reverse sampling loop, and a smoke test that verifies noise addition, model output, and sampling shapes.Sun, 21 Sep 2025 00:00:00 GMThttps://ailinkdeeptech.com/research/pagedattention/https://ailinkdeeptech.com/research/pagedattention/Master PagedAttention: virtual memory KV cache paging, block tables, prefix caching (CoW), continuous batching, memory fragmentation math, and PyTorch kernels.Thu, 18 Sep 2025 00:00:00 GMThttps://ailinkdeeptech.com/research/deepseek_algo/https://ailinkdeeptech.com/research/deepseek_algo/A PyTorch implementation of the DeepSeek-style architecture: RMSNorm pre-norm blocks, a multi-head latent attention module, a Mixture-of-Experts (MoE) feed-forward router, and a stack of DeepSeek blocks with token-level residual paths.Tue, 09 Sep 2025 00:00:00 GMThttps://ailinkdeeptech.com/research/ddim/https://ailinkdeeptech.com/research/ddim/Master DDIM (Denoising Diffusion Implicit Models): non-Markovian sampling, probability-flow ODE derivations, deterministic inference, and PyTorch pipelines.Fri, 05 Sep 2025 00:00:00 GMThttps://ailinkdeeptech.com/research/visiontransformer_algo/https://ailinkdeeptech.com/research/visiontransformer_algo/A PyTorch implementation of the Vision Transformer (ViT) with patch embedding, learnable class token and position embeddings, multi-head self-attention, MLP blocks, and classification head.Thu, 04 Sep 2025 00:00:00 GMThttps://ailinkdeeptech.com/research/vision_transformer/https://ailinkdeeptech.com/research/vision_transformer/Master the Vision Transformer (ViT): patch projection, class tokens, multi-head self-attention over patches, DeiT augmentation, and PyTorch implementations.Mon, 01 Sep 2025 00:00:00 GMThttps://ailinkdeeptech.com/research/llama_algo/https://ailinkdeeptech.com/research/llama_algo/A from-scratch PyTorch implementation of the LLaMA-style Transformer: RMSNorm, Rotary Position Embeddings (RoPE), GQA-style multi-head attention with causal mask, SwiGLU MLP, and a stack of pre-norm decoder layers.Wed, 27 Aug 2025 00:00:00 GMThttps://ailinkdeeptech.com/research/rope/https://ailinkdeeptech.com/research/rope/Master RoPE (Rotary Position Embedding): complex rotation algebra, relative distance invariance proof, long-context scaling (YaRN), and PyTorch kernels.Mon, 11 Aug 2025 00:00:00 GMThttps://ailinkdeeptech.com/research/ddpm/https://ailinkdeeptech.com/research/ddpm/Master DDPM: forward and reverse Gaussian Markov chains, closed-form ELBO derivation, noise-prediction parameterization, and PyTorch training pipelines.Tue, 05 Aug 2025 00:00:00 GMThttps://ailinkdeeptech.com/research/3dgs_algo/https://ailinkdeeptech.com/research/3dgs_algo/A PyTorch implementation of 3D Gaussian Splatting from scratch: per-Gaussian learnable positions, scales, quaternions and RGBA, 3D covariance via R S Sᵀᵀ, projection to 2D via the perspective Jacobian, splat-based alpha compositing over a pixel grid, and a smoke test that checks initialization, symmetry/orthogonality of covariances, and the rendering output shape and value range.Mon, 04 Aug 2025 00:00:00 GMThttps://ailinkdeeptech.com/research/diffusion/https://ailinkdeeptech.com/research/diffusion/Master diffusion models: forward Markov perturbations, reverse ELBO derivations, score-based SDE unifications, and production PyTorch training pipelines.Fri, 01 Aug 2025 00:00:00 GMThttps://ailinkdeeptech.com/research/a3c_algo/https://ailinkdeeptech.com/research/a3c_algo/A PyTorch implementation of Asynchronous Advantage Actor-Critic (A3C): a shared Actor-Critic network with a softmax policy head, multiple Worker processes that collect CartPole trajectories, compute n-step returns and advantages, and sync gradients into a global model, plus a greedy test harness that runs the trained policy in render mode.Sun, 20 Jul 2025 00:00:00 GMThttps://ailinkdeeptech.com/research/4dgs/https://ailinkdeeptech.com/research/4dgs/Master 4D Gaussian Splatting (4DGS): HexPlane spatiotemporal neural fields, canonical deformation fields, and real-time dynamic novel view synthesis.Thu, 10 Jul 2025 12:37:00 GMThttps://ailinkdeeptech.com/research/3dgs/https://ailinkdeeptech.com/research/3dgs/Master 3D Gaussian Splatting (3DGS): covariance matrix decomposition, tile-based differentiable rasterization, adaptive density control, and real-time CUDA pipelines.Fri, 04 Jul 2025 12:35:00 GMThttps://ailinkdeeptech.com/research/llava/https://ailinkdeeptech.com/research/llava/Master LLaVA: visual instruction tuning, 2-stage cross-modal projection, CLIP/SigLIP feature alignment, AnyRes scaling, and PyTorch VLM pipelines.Tue, 10 Jun 2025 00:00:00 GMThttps://ailinkdeeptech.com/research/stable_diffusion/https://ailinkdeeptech.com/research/stable_diffusion/Master Stable Diffusion: latent diffusion models (LDM), perceptual VAE compression, cross-attention U-Net, classifier-free guidance, and PyTorch pipelines.Sun, 25 May 2025 00:00:00 GMThttps://ailinkdeeptech.com/research/lora/https://ailinkdeeptech.com/research/lora/Master LoRA (Low-Rank Adaptation): low-rank matrix decomposition, intrinsic rank analysis, QLoRA NF4 quantization, DoRA, and PyTorch PEFT pipelines.Mon, 21 Apr 2025 00:00:00 GMThttps://ailinkdeeptech.com/research/controlnet/https://ailinkdeeptech.com/research/controlnet/Master ControlNet: zero-convolution weight initialization, trainable U-Net copy architecture, multi-condition composition, and PyTorch training pipelines.Sat, 05 Apr 2025 00:00:00 GMThttps://ailinkdeeptech.com/research/transformer3d/https://ailinkdeeptech.com/research/transformer3d/Master 3D Transformers: vector self-attention on point clouds, sparse voxel hashing, 3D Swin windowing, relative positional bias, and PyTorch architectures.Fri, 04 Apr 2025 00:00:00 GMThttps://ailinkdeeptech.com/research/llama/https://ailinkdeeptech.com/research/llama/Master Llama architecture: RMSNorm pre-normalization, SwiGLU activations, RoPE relative embeddings, Grouped-Query Attention (GQA), and PyTorch pipelines.Sat, 15 Mar 2025 00:00:00 GMThttps://ailinkdeeptech.com/research/transformer/https://ailinkdeeptech.com/research/transformer/Master the Transformer: scaled dot-product attention, multi-head projections, sinusoidal & RoPE encodings, Pre-LN vs Post-LN stability, and PyTorch engines.Mon, 10 Feb 2025 00:00:00 GMT