docs/source/en/internal/rope_utils.md
This page explains how the Rotary Embedding is computed and applied in Transformers and what types of RoPE are supported.
Rotary Position Embeddings are a technique used to inject positional information into attention mechanisms without relying on explicit position encodings.
Instead of adding position vectors to token embeddings, RoPE rotates query and key vectors in the complex plane according to their positions enabling relative positional awareness and better extrapolation to unseen sequence lengths.
The Transformers library provides a flexible and extensible implementation of various RoPE types defined in [~modeling_rope_utils.ROPE_VALIDATION_FUNCTIONS], including both the default and scaled variants:
| Rope Type | Description |
|---|---|
"default" | Standard rotary embedding as in LLaMA. |
"linear" | Linear-scaled RoPE which allows longer context windows. |
"dynamic" | NTK-aware scaling computed by rescaling frequency base (θ) for longer context. |
"yarn" | YaRN scaling variant providing smoother extrapolation and stability. |
"longrope" | LongRoPE scaling as in Phi-2 model series. |
"llama3" | RoPE scaling as in Llama3.1. |
To enable and customize rotary embeddings, add a rope_parameters field to your model’s configuration file (config.json). This field controls the RoPE behavior across model layers. Note that each RoPE variant defines its own set of expected keys and missing keys will raise an error. See the example below which creates a llama config with default RoPE parameters:
from transformers import LlamaConfig
config = LlamaConfig()
config.rope_parameters = {
"rope_type": "default", # type of RoPE to use
# rope_theta is optional — omitting it uses the model’s default_theta (typically 10000.0)
}
# If we want to apply a scaled RoPE type, we need to pass extra parameters
config.rope_parameters = {
"rope_type": "linear",
"rope_theta": 10000.0, # can be omitted to fall back to default_theta
"factor": 8.0 # scale factor for context extension
}
Some models such as Gemma-3 use different layer types with different attention mechanisms, i.e. "full attention" in some blocks and "sliding-window attention" in others. Transformers supports specifying distinct RoPE parameters per layer type for these models. In this case, rope_parameters should be a nested dictionary, where top-level keys correspond to config.layer_types and values are per-type RoPE parameters. During model initialization, each decoder layer will automatically look up the matching RoPE configuration based on its declared layer type.
from transformers import Gemma3Config
config = Gemma3Config()
config.rope_parameters = {
"full_attention": {
"rope_type": "dynamic",
"rope_theta": 1000000.0,
"factor": 8.0,
"original_max_position_embeddings": 8096,
},
"sliding_attention": {
"rope_type": "default",
"rope_theta": 10000.0,
}
}
[[autodoc]] RopeParameters - call