Skip to content

[Bugfix] Fix Plamo3 rope handling #29092

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
DarkLight1337 merged 4 commits into from
Nov 21, 2025
Merged

[Bugfix] Fix Plamo3 rope handling #29092

Changes from all commits
Commits
Show all changes
4 commits
Select commit Hold shift + click to select a range
2fd7584
[Bugfix] Fix Plamo3 rope handling
DarkLight1337 Nov 20, 2025
5b6bba3
Fix
DarkLight1337 Nov 20, 2025
b44c6de
Unnecessary attribute
DarkLight1337 Nov 20, 2025
03a16e5
Make sure that rope parameters for sliding is truly default
hmellor Nov 20, 2025
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
26 changes: 18 additions & 8 deletions vllm/model_executor/models/plamo3.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -62,7 +62,7 @@ class Plamo3Config(PretrainedConfig): # type: ignore
# if `sliding_window` is list
interleaved_sliding_window: list[int | None]
sliding_window_pattern: int
rope_theta: int
rope_parameters: dict[str, Any]
rope_local_theta: int
# MLP
intermediate_size: int
Expand Down Expand Up @@ -153,13 +153,24 @@ def __init__(self, *, vllm_config: VllmConfig, prefix: str = "", **kwargs) -> No
quant_config=quant_config,
prefix=f"{prefix}.o_proj",
)

layer_idx = extract_layer_index(prefix)
full_attn = config.interleaved_sliding_window[layer_idx] is None
layer_type = config.layer_types[layer_idx]
is_sliding = layer_type == "sliding_attention"

self.rope_theta = config.rope_theta if full_attn else config.rope_local_theta
self.rope_scaling = (
config.rope_scaling if hasattr(config, "rope_scaling") else None
)
# Initialize the rotary embedding.
if layer_type in config.rope_parameters:
# Transformers v5 rope config.
rope_parameters = config.rope_parameters[layer_type]
else:
# Transformers v4 rope config.
# Global attention. Use the values in config.json.
rope_parameters = config.rope_parameters
# Local attention. Override the values in config.json.
if is_sliding:
rope_parameters = dict(
rope_type="default", rope_theta=config.rope_local_theta
)
max_position = config.max_position_embeddings
if hasattr(vllm_config.model_config, "max_model_len") and isinstance(
vllm_config.model_config.max_model_len, int
Expand All @@ -170,8 +181,7 @@ def __init__(self, *, vllm_config: VllmConfig, prefix: str = "", **kwargs) -> No
self.head_dim,
rotary_dim=self.head_dim,
max_position=max_position,
base=self.rope_theta,
rope_scaling=self.rope_scaling,
rope_parameters=rope_parameters,
)
self.q_norm = RMSNorm(self.head_dim, eps=config.rms_norm_eps)
set_weight_attrs(
Expand Down