Add pydocstyle to the ruff rules (#52136)

Signed-off-by: Harry Mellor <19981378+hmellor@users.noreply.github.com>
Signed-off-by: Harry Mellor <hej.mellor@gmail.com>
Co-authored-by: Claude Opus 5 (1M context) <noreply@anthropic.com>
This commit is contained in:
Harry Mellor
2026-09-16 16:44:29 +00:00
committed by GitHub
co-authored by Claude Opus 5
parent 804e5377aa
commit 8c1557a79c
2002 changed files with 15024 additions and 14269 deletions
+31 -2
View File
@@ -71,6 +71,8 @@ class Zamba2LoRA(nn.Module):
rank: LoRA rank
output_dim: output dimension
quant_config: Configuration for model quantization
prefix: Module prefix used for quantization config lookup
"""
super().__init__()
@@ -130,6 +132,7 @@ class Zamba2Attention(nn.Module):
cache_config: Configuration for key-value caching
quant_config: Configuration for model quantization
prefix: Optional prefix for parameter names
"""
super().__init__()
tp_size = get_tensor_model_parallel_world_size()
@@ -257,6 +260,7 @@ class Zamba2Attention(nn.Module):
Returns:
Output tensor [batch_size, seq_len, hidden_size]
"""
qkv, _ = self.qkv_proj(hidden_states)
query_states, key_states, value_states = qkv.split([self.qkv_size] * 3, dim=-1)
@@ -311,6 +315,8 @@ class Zamba2MLP(nn.Module):
bare_block_idx: Index of the bare block in the model
num_hybrid_layers: Total number of hybrid layers
quant_config: Configuration for model quantization
prefix: Module prefix used for quantization config lookup
"""
super().__init__()
self.config = config
@@ -369,6 +375,7 @@ class Zamba2MLP(nn.Module):
Returns:
Output tensor [batch_size, seq_len, hidden_size] after applying
gated feed-forward transformation
"""
# Project input to intermediate size with gating
gate_up_states, _ = self.gate_up_proj(hidden_states)
@@ -415,6 +422,7 @@ class Zamba2AttentionDecoderLayer(nn.Module):
cache_config: Configuration for key-value caching
quant_config: Configuration for model quantization
prefix: Optional prefix for parameter names
"""
super().__init__()
@@ -461,8 +469,8 @@ class Zamba2AttentionDecoderLayer(nn.Module):
Returns:
Transformed hidden states after attention and feed-forward
"""
"""
# The argument original_hidden_states is concatenated with hidden_states
# (which is the output of the previous (mamba) layer).
# The concatenated tensor is then used as input of the pre-attention
@@ -510,7 +518,11 @@ class Zamba2MambaDecoderLayer(nn.Module):
Args:
config: The Zamba2 model configuration
model_config: The model config, when available
cache_config: The KV cache config, when available
quant_config: Configuration for model quantization
prefix: Module prefix used for quantization config lookup
"""
super().__init__()
@@ -555,6 +567,7 @@ class Zamba2MambaDecoderLayer(nn.Module):
Returns:
Transformed hidden states with residual connection applied
"""
# Store input for residual connection
residual = hidden_states
@@ -603,6 +616,13 @@ class Zamba2HybridLayer(nn.Module):
Args:
shared_transformer: Transformer decoder layer for attention pathway
config: The Zamba2 model configuration
block_idx: Index of this hybrid block in the model
model_config: The model config, when available
cache_config: The KV cache config, when available
quant_config: Configuration for model quantization
prefix: Module prefix used for quantization config lookup
"""
super().__init__()
self.block_idx = block_idx
@@ -644,6 +664,7 @@ class Zamba2HybridLayer(nn.Module):
Returns:
Output tensor combining transformer and Mamba representations
"""
# Process through transformer pathway
transformer_hidden_states = self.shared_transformer(
@@ -680,6 +701,7 @@ class Zamba2Model(nn.Module):
vllm_config: Configuration object containing model, cache,
quantization and LoRA settings
prefix: Optional prefix for parameter names in state dict
"""
super().__init__()
@@ -765,6 +787,7 @@ class Zamba2Model(nn.Module):
Returns:
Embedded representation of the input tokens
"""
return self.embed_tokens(input_ids)
@@ -784,6 +807,7 @@ class Zamba2Model(nn.Module):
Returns:
Either final hidden states or intermediate tensors for pipeline
parallelism
"""
# Handle pipeline parallelism for first rank
if inputs_embeds is None:
@@ -853,8 +877,8 @@ class Zamba2ForCausalLM(nn.Module, HasInnerState, IsHybrid, SupportsMambaPrefixC
Tuple containing:
- conv_state_shape: Shape for convolutional state cache
- temporal_state_shape: Shape for state space model cache
"""
"""
parallel_config = vllm_config.parallel_config
hf_config = vllm_config.model_config.hf_config
intermediate_size = hf_config.mamba_expand * hf_config.hidden_size
@@ -884,6 +908,7 @@ class Zamba2ForCausalLM(nn.Module, HasInnerState, IsHybrid, SupportsMambaPrefixC
Raises:
AssertionError: If prefix caching is enabled
(not supported by Mamba)
"""
config = vllm_config.model_config.hf_config
@@ -914,10 +939,12 @@ class Zamba2ForCausalLM(nn.Module, HasInnerState, IsHybrid, SupportsMambaPrefixC
def embed_input_ids(self, input_ids: torch.Tensor) -> torch.Tensor:
"""Convert input token IDs to embeddings.
Args:
input_ids: Tensor of input token IDs
Returns:
Embedded representation of the input tokens
"""
return self.model.embed_input_ids(input_ids)
@@ -938,6 +965,7 @@ class Zamba2ForCausalLM(nn.Module, HasInnerState, IsHybrid, SupportsMambaPrefixC
Returns:
Output hidden states
"""
# Forward pass through model
hidden_states = self.model(
@@ -959,6 +987,7 @@ class Zamba2ForCausalLM(nn.Module, HasInnerState, IsHybrid, SupportsMambaPrefixC
Returns:
Logits for next token prediction
"""
logits = self.logits_processor(self.lm_head, hidden_states)
return logits