# gm.nn

[[[Source]]](https://github.com/google-deepmind/gemma/tree/main/gemma/gm/nn/__init__.py)

```{eval-rst}
.. automodule:: gemma.gm.nn
  :no-members:
```

## Symbols


### Module

|  |  |
--- | ---
[gm.nn.config](config/index) | Symbols needed to build new `TransformerConfig`.

### Class

|  |  |
--- | ---
[gm.nn.AnchoredPolicy](AnchoredPolicy) | Wrapper around a model to compute policy and anchor outputs.
[gm.nn.AnchoredPolicyOutput](AnchoredPolicyOutput) | Output of the `gm.nn.AnchoredPolicy`.
[gm.nn.Attention](Attention) | Attention module.
[gm.nn.AttentionType](AttentionType) | 
[gm.nn.Block](Block) | Transformer block.
[gm.nn.Einsum](Einsum) | Einsum is a convenience module for parameterized tensor multiplication.
[gm.nn.Embedder](Embedder) | Embedder module.
[gm.nn.FeedForward](FeedForward) | Feed forward module.
[gm.nn.Gemma2_27B](Gemma2_27B) | Gemma2 transformer architecture.
[gm.nn.Gemma2_2B](Gemma2_2B) | Gemma2 transformer architecture.
[gm.nn.Gemma2_9B](Gemma2_9B) | Gemma2 transformer architecture.
[gm.nn.Gemma3_12B](Gemma3_12B) | Gemma3 transformer architecture.
[gm.nn.Gemma3_1B](Gemma3_1B) | Gemma3 transformer architecture.
[gm.nn.Gemma3_270M](Gemma3_270M) | Gemma3 transformer architecture.
[gm.nn.Gemma3_27B](Gemma3_27B) | Gemma3 transformer architecture.
[gm.nn.Gemma3_4B](Gemma3_4B) | Gemma3 transformer architecture.
[gm.nn.Gemma3n_E2B](Gemma3n_E2B) | Gemma3n E2B transformer architecture.
[gm.nn.Gemma3n_E4B](Gemma3n_E4B) | Gemma3n E4B transformer architecture.
[gm.nn.Gemma4_26B_A4B](Gemma4_26B_A4B) | Gemma 4 26B_A4B MoE model.
[gm.nn.Gemma4_31B](Gemma4_31B) | Gemma 4 31B model.
[gm.nn.Gemma4_E2B](Gemma4_E2B) | Gemma 4 E2B model.
[gm.nn.Gemma4_E4B](Gemma4_E4B) | Gemma 4 E4B model.
[gm.nn.IntWrapper](IntWrapper) | Wrapper around a Gemma model to enable int4 inference.
[gm.nn.LoRA](LoRA) | Wrapper around a Gemma model to enable LoRA.
[gm.nn.Output](Output) | Output of the Gemma model.
[gm.nn.QuantizationAwareWrapper](QuantizationAwareWrapper) | Wrapper around a Gemma model to enable quantization aware training.
[gm.nn.RMSNorm](RMSNorm) | RMSNorm layer.
[gm.nn.SigLiPFromPatches](SigLiPFromPatches) | SigLIP vision encoder forward pass from PatchifiedMedia.
[gm.nn.Transformer](Transformer) | Base transformer class.
[gm.nn.TransformerLike](TransformerLike) | Protocol for a transformer model to be used with a Sampler.

### Typing

|  |  |
--- | ---
[gm.nn.Cache](Cache) | 

```{toctree}
:hidden:

AnchoredPolicy
AnchoredPolicyOutput
Attention
AttentionType
Block
Cache
Einsum
Embedder
FeedForward
Gemma2_27B
Gemma2_2B
Gemma2_9B
Gemma3_12B
Gemma3_1B
Gemma3_270M
Gemma3_27B
Gemma3_4B
Gemma3n_E2B
Gemma3n_E4B
Gemma4_26B_A4B
Gemma4_31B
Gemma4_E2B
Gemma4_E4B
IntWrapper
LoRA
Output
QuantizationAwareWrapper
RMSNorm
SigLiPFromPatches
Transformer
TransformerLike
config/index
```