# gizml

Tensors on the CPU and GPU, gradients, GPU kernels, and language, image and video models. Import a module by name, for example `import * as Store from "gizdb:store";`. Each page lists the module's public declarations exactly as an installed toolchain provides them.

| Module | Purpose |
| --- | --- |
| [gizml:ai](/docs/reference/gizml/ai) | AI models on this computer, by name. |
| [gizml:cluster](/docs/reference/gizml/cluster) | clustering with a predict step: k-means (k-means++ and Lloyd), mini-batch k-means, BIRCH (CF subclusters, then Ward agglomeration by the nearest-neighbor chain), affinity propagation and mean shift. |
| [gizml:diffusion](/docs/reference/gizml/diffusion) | video and image generation models (diffusion transformers), fast on any hardware.  A model family is a recipe: its config.json read as an Arch, and the names of its tensors. |
| [gizml:grad](/docs/reference/gizml/grad) | derivatives of tensor programs. |
| [gizml:grammar](/docs/reference/gizml/grammar) | the texts a model may write, for constrained decoding. |
| [gizml:llm](/docs/reference/gizml/llm) | language models, fast on any hardware.  A model family is a recipe: its config.json read as an Arch, and the names of its tensors. |
| [gizml:ltx](/docs/reference/gizml/ltx) | Lightricks' LTX-2.5 (22B), the audio and video flow transformer of giz.ai's ltx2-5-distilled.  Two streams of rows run side by side through 48 blocks: video (4096 wide, 32 heads of 128; a row a latent pixel of 8 frames and 32 x 32 pixels) and audio (2048 wide, 32 heads of 64; a row a 40 ms latent). |
| [gizml:ml](/docs/reference/gizml/ml) | one interface over the estimators of gizlang:{linalg,trees,svm,cluster}, with the algorithms and default hyperparameters of scikit-learn, xgboost, LightGBM and CatBoost: fit, predict, predict_proba, the metrics, and the model bytes. |
| [gizml:netops](/docs/reference/gizml/netops) | the operations of networks on gizml:tensor Tensors. |
| [gizml:nn](/docs/reference/gizml/nn) | what training a network needs on top of gizml:grad. |
| [gizml:npullm](/docs/reference/gizml/npullm) | Gemma 4 E2B's forward pass on an AMD NPU (XDNA 2), with the AIE kernels of FastFlowLM (the layer and lm_head xclbins, MIT runtime, binaries free to use unmodified). |
| [gizml:onnx](/docs/reference/gizml/onnx) | ONNX model files (ModelProto, IR version 8) of gizml:ml models, and gizml:ml models of ONNX files. |
| [gizml:quant](/docs/reference/gizml/quant) | the formats of stored weights, apart from the model families and the kernels that run them. |
| [gizml:svm](/docs/reference/gizml/svm) | support vector machines. |
| [gizml:tensor](/docs/reference/gizml/tensor) | n-dimensional f32 arrays on the CPU's vector lanes and cores. |
| [gizml:tile](/docs/reference/gizml/tile) | GPU kernels as ordinary Gizlang over tiles, on every device.  A tile program says what one block of the output is; this module decides everything else: threads, which thread holds which element, workgroup memory, barriers, bounds masks, matrix-core fragments, layout conversions and the device. |
| [gizml:tokenizer](/docs/reference/gizml/tokenizer) | text into model inputs. |
| [gizml:trees](/docs/reference/gizml/trees) | decision trees and their ensembles: CART (scikit-learn's exact splitter and its random splitter for extra trees), forests, bagging, AdaBoost (SAMME and R2), gradient boosting, and one histogram GBDT with depth-wise, leaf-wise and oblivious growth that the xgboost, LightGBM, HistGradientBoosting and CatBoost defaults share. |
| [gizml:video](/docs/reference/gizml/video) | video on top of gizhw:codec and gizlang:mp4. |

gizml

Tensors on the CPU and GPU, gradients, GPU kernels, and language, image and video models. Import a module by name, for example import * as Store from "gizdb:store";. Each page lists the module's public declarations exactly as an installed toolchain provides them.

Module	Purpose
gizml:ai	AI models on this computer, by name.
gizml:cluster	clustering with a predict step: k-means (k-means++ and Lloyd), mini-batch k-means, BIRCH (CF subclusters, then Ward agglomeration by the nearest-neighbor chain), affinity propagation and mean shift.
gizml:diffusion	video and image generation models (diffusion transformers), fast on any hardware. A model family is a recipe: its config.json read as an Arch, and the names of its tensors.
gizml:grad	derivatives of tensor programs.
gizml:grammar	the texts a model may write, for constrained decoding.
gizml:llm	language models, fast on any hardware. A model family is a recipe: its config.json read as an Arch, and the names of its tensors.
gizml:ltx	Lightricks' LTX-2.5 (22B), the audio and video flow transformer of giz.ai's ltx2-5-distilled. Two streams of rows run side by side through 48 blocks: video (4096 wide, 32 heads of 128; a row a latent pixel of 8 frames and 32 x 32 pixels) and audio (2048 wide, 32 heads of 64; a row a 40 ms latent).
gizml:ml	one interface over the estimators of gizlang:{linalg,trees,svm,cluster}, with the algorithms and default hyperparameters of scikit-learn, xgboost, LightGBM and CatBoost: fit, predict, predict_proba, the metrics, and the model bytes.
gizml:netops	the operations of networks on gizml:tensor Tensors.
gizml:nn	what training a network needs on top of gizml:grad.
gizml:npullm	Gemma 4 E2B's forward pass on an AMD NPU (XDNA 2), with the AIE kernels of FastFlowLM (the layer and lm_head xclbins, MIT runtime, binaries free to use unmodified).
gizml:onnx	ONNX model files (ModelProto, IR version 8) of gizml:ml models, and gizml:ml models of ONNX files.
gizml:quant	the formats of stored weights, apart from the model families and the kernels that run them.
gizml:svm	support vector machines.
gizml:tensor	n-dimensional f32 arrays on the CPU's vector lanes and cores.
gizml:tile	GPU kernels as ordinary Gizlang over tiles, on every device. A tile program says what one block of the output is; this module decides everything else: threads, which thread holds which element, workgroup memory, barriers, bounds masks, matrix-core fragments, layout conversions and the device.
gizml:tokenizer	text into model inputs.
gizml:trees	decision trees and their ensembles: CART (scikit-learn's exact splitter and its random splitter for extra trees), forests, bagging, AdaBoost (SAMME and R2), gradient boosting, and one histogram GBDT with depth-wise, leaf-wise and oblivious growth that the xgboost, LightGBM, HistGradientBoosting and CatBoost defaults share.
gizml:video	video on top of gizhw:codec and gizlang:mp4.