Skip to content

vllm.model_executor.models.gemma3n

logger module-attribute 

logger = init_logger(__name__)

Gemma3nAltUp

Bases: Module

Alternating updates (Altup) The AltUp module wraps transformer layers. The predict step modifies the input to the transformer layer, and the correct step propagates the output of the transformer layer to the sparsely updated dimensions. See more in the research paper: https://proceedings.neurips.cc/paper_files/paper/2023/file/f2059277ac6ce66e7e5543001afa8bb5-Paper-Conference.pdf

Source code in vllm/model_executor/models/gemma3n.py 
 55
 56
 57
 58
 59
 60
 61
 62
 63
 64
 65
 66
 67
 68
 69
 70
 71
 72
 73
 74
 75
 76
 77
 78
 79
 80
 81
 82
 83
 84
 85
 86
 87
 88
 89
 90
 91
 92
 93
 94
 95
 96
 97
 98
 99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
class Gemma3nAltUp(nn.Module):
    """Alternating updates (Altup)
    The AltUp module wraps transformer layers. The `predict` step modifies the
    input to the transformer layer, and the `correct` step propagates the output
    of the transformer layer to the sparsely updated dimensions.
    See more in the research paper:
    https://proceedings.neurips.cc/paper_files/paper/2023/file/f2059277ac6ce66e7e5543001afa8bb5-Paper-Conference.pdf
    """

    def __init__(
        self,
        hidden_size: int,
        rms_norm_eps: float,
        altup_num_inputs: int,
        altup_coef_clip: float,
        altup_active_idx: int,
        prefix: str,
    ):
        super().__init__()

        self.altup_num_inputs = altup_num_inputs
        self.altup_active_idx = altup_active_idx
        self.altup_coef_clip = altup_coef_clip

        self.correction_coefs = ReplicatedLinear(
            altup_num_inputs,
            altup_num_inputs,
            bias=False,
            prefix=f"{prefix}.correction_coefs",
            return_bias=False,
        )
        self.prediction_coefs = ReplicatedLinear(
            altup_num_inputs,
            altup_num_inputs**2,
            bias=False,
            prefix=f"{prefix}.prediction_coefs",
            return_bias=False,
        )
        self.modality_router = ReplicatedLinear(
            hidden_size,
            altup_num_inputs,
            bias=False,
            prefix=f"{prefix}.modality_router",
            return_bias=False,
        )
        self.router_norm = RMSNorm(
            hidden_size=hidden_size,
            eps=rms_norm_eps,
        )
        self.router_input_scale = torch.tensor(
            hidden_size**-1.0, dtype=self.modality_router.weight.dtype)
        self.correct_output_scale = nn.Parameter(
            torch.zeros(hidden_size, dtype=torch.float32))

    def _compute_router_modalities(self, x: torch.Tensor) -> torch.Tensor:
        router_inputs = self.router_norm(x) * self.router_input_scale
        routed = self.modality_router(router_inputs)
        return torch.tanh(routed.float()).type_as(x)

    def scale_corrected_output(self, corrected: torch.Tensor) -> torch.Tensor:
        return (corrected.type_as(self.correct_output_scale) *
                self.correct_output_scale).type_as(corrected)

    def predict(self, hidden_states: torch.Tensor) -> torch.Tensor:
        # hidden:       [altup_num_inputs, num_tokens, hidden_size]
        # modalities:   [num_tokens, num_altup_inputs]
        # all_coefs:    [num_tokens, num_altup_inputs ** 2]
        modalities = self._compute_router_modalities(
            hidden_states[self.altup_active_idx])
        all_coefs = self.prediction_coefs(modalities)

        # Reshape and transpose the 2D matrix for the matmul.
        # all_coefs_T:  [num_tokens, num_altup_inputs, num_altup_inputs]
        all_coefs_T = all_coefs.reshape(
            -1,
            self.altup_num_inputs,
            self.altup_num_inputs,
        ).permute(0, 2, 1)

        # hidden_states to [num_tokens, hidden_size, altup_num_inputs]
        predictions = torch.matmul(hidden_states.permute(1, 2, 0), all_coefs_T)
        # [altup_num_inputs, num_tokens, hidden_size]
        predictions = predictions.permute(2, 0, 1)
        predictions += hidden_states
        return predictions.contiguous()

    def correct(self, predictions: torch.Tensor,
                activated: torch.Tensor) -> torch.Tensor:
        # predictions:  [altup_num_inputs, num_tokens, hidden_size]
        # activated:    [num_tokens, hidden_size]
        # modalities:   [num_tokens, altup_num_inputs]
        modalities = self._compute_router_modalities(activated)
        # innovation:   [num_tokens, altup_num_inputs]
        innovation = activated - predictions[self.altup_active_idx]
        # innovation:   [altup_num_inputs, num_tokens, hidden_size]
        innovation = innovation.repeat(self.altup_num_inputs, 1, 1)

        # Permute to [altup_num_inputs, num_tokens] as the last dim
        # is a scalar applied to each altup input and expand on
        # num_tokens dim for broadcastability over hidden_size.
        # all_coefs:    [num_tokens, altup_num_inputs]
        all_coefs = self.correction_coefs(modalities) + 1.0
        # all_coefs:    [altup_num_inputs, num_tokens, 1]
        all_coefs = all_coefs.T.unsqueeze(-1)

        # Elementwise (broadcast over hidden_size).
        corrected = torch.mul(innovation, all_coefs)
        corrected += predictions

        return corrected.contiguous()

altup_active_idx instance-attribute 

altup_active_idx = altup_active_idx

altup_coef_clip instance-attribute 

altup_coef_clip = altup_coef_clip

altup_num_inputs instance-attribute 

altup_num_inputs = altup_num_inputs

correct_output_scale instance-attribute 

correct_output_scale = Parameter(
    zeros(hidden_size, dtype=float32)
)

correction_coefs instance-attribute 

correction_coefs = ReplicatedLinear(
    altup_num_inputs,
    altup_num_inputs,
    bias=False,
    prefix=f"{prefix}.correction_coefs",
    return_bias=False,
)

modality_router instance-attribute 

modality_router = ReplicatedLinear(
    hidden_size,
    altup_num_inputs,
    bias=False,
    prefix=f"{prefix}.modality_router",
    return_bias=False,
)

prediction_coefs instance-attribute 

prediction_coefs = ReplicatedLinear(
    altup_num_inputs,
    altup_num_inputs**2,
    bias=False,
    prefix=f"{prefix}.prediction_coefs",
    return_bias=False,
)

router_input_scale instance-attribute 

router_input_scale = tensor(hidden_size**-1.0, dtype=dtype)

router_norm instance-attribute 

router_norm = RMSNorm(
    hidden_size=hidden_size, eps=rms_norm_eps
)

__init__ 

__init__(
    hidden_size: int,
    rms_norm_eps: float,
    altup_num_inputs: int,
    altup_coef_clip: float,
    altup_active_idx: int,
    prefix: str,
)
Source code in vllm/model_executor/models/gemma3n.py 
 64
 65
 66
 67
 68
 69
 70
 71
 72
 73
 74
 75
 76
 77
 78
 79
 80
 81
 82
 83
 84
 85
 86
 87
 88
 89
 90
 91
 92
 93
 94
 95
 96
 97
 98
 99
100
101
102
103
104
105
106
107
def __init__(
    self,
    hidden_size: int,
    rms_norm_eps: float,
    altup_num_inputs: int,
    altup_coef_clip: float,
    altup_active_idx: int,
    prefix: str,
):
    super().__init__()

    self.altup_num_inputs = altup_num_inputs
    self.altup_active_idx = altup_active_idx
    self.altup_coef_clip = altup_coef_clip

    self.correction_coefs = ReplicatedLinear(
        altup_num_inputs,
        altup_num_inputs,
        bias=False,
        prefix=f"{prefix}.correction_coefs",
        return_bias=False,
    )
    self.prediction_coefs = ReplicatedLinear(
        altup_num_inputs,
        altup_num_inputs**2,
        bias=False,
        prefix=f"{prefix}.prediction_coefs",
        return_bias=False,
    )
    self.modality_router = ReplicatedLinear(
        hidden_size,
        altup_num_inputs,
        bias=False,
        prefix=f"{prefix}.modality_router",
        return_bias=False,
    )
    self.router_norm = RMSNorm(
        hidden_size=hidden_size,
        eps=rms_norm_eps,
    )
    self.router_input_scale = torch.tensor(
        hidden_size**-1.0, dtype=self.modality_router.weight.dtype)
    self.correct_output_scale = nn.Parameter(
        torch.zeros(hidden_size, dtype=torch.float32))

_compute_router_modalities 

_compute_router_modalities(x: Tensor) -> Tensor
Source code in vllm/model_executor/models/gemma3n.py 
109
110
111
112
def _compute_router_modalities(self, x: torch.Tensor) -> torch.Tensor:
    router_inputs = self.router_norm(x) * self.router_input_scale
    routed = self.modality_router(router_inputs)
    return torch.tanh(routed.float()).type_as(x)

correct 

correct(predictions: Tensor, activated: Tensor) -> Tensor
Source code in vllm/model_executor/models/gemma3n.py 
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
def correct(self, predictions: torch.Tensor,
            activated: torch.Tensor) -> torch.Tensor:
    # predictions:  [altup_num_inputs, num_tokens, hidden_size]
    # activated:    [num_tokens, hidden_size]
    # modalities:   [num_tokens, altup_num_inputs]
    modalities = self._compute_router_modalities(activated)
    # innovation:   [num_tokens, altup_num_inputs]
    innovation = activated - predictions[self.altup_active_idx]
    # innovation:   [altup_num_inputs, num_tokens, hidden_size]
    innovation = innovation.repeat(self.altup_num_inputs, 1, 1)

    # Permute to [altup_num_inputs, num_tokens] as the last dim
    # is a scalar applied to each altup input and expand on
    # num_tokens dim for broadcastability over hidden_size.
    # all_coefs:    [num_tokens, altup_num_inputs]
    all_coefs = self.correction_coefs(modalities) + 1.0
    # all_coefs:    [altup_num_inputs, num_tokens, 1]
    all_coefs = all_coefs.T.unsqueeze(-1)

    # Elementwise (broadcast over hidden_size).
    corrected = torch.mul(innovation, all_coefs)
    corrected += predictions

    return corrected.contiguous()

predict 

predict(hidden_states: Tensor) -> Tensor
Source code in vllm/model_executor/models/gemma3n.py 
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
def predict(self, hidden_states: torch.Tensor) -> torch.Tensor:
    # hidden:       [altup_num_inputs, num_tokens, hidden_size]
    # modalities:   [num_tokens, num_altup_inputs]
    # all_coefs:    [num_tokens, num_altup_inputs ** 2]
    modalities = self._compute_router_modalities(
        hidden_states[self.altup_active_idx])
    all_coefs = self.prediction_coefs(modalities)

    # Reshape and transpose the 2D matrix for the matmul.
    # all_coefs_T:  [num_tokens, num_altup_inputs, num_altup_inputs]
    all_coefs_T = all_coefs.reshape(
        -1,
        self.altup_num_inputs,
        self.altup_num_inputs,
    ).permute(0, 2, 1)

    # hidden_states to [num_tokens, hidden_size, altup_num_inputs]
    predictions = torch.matmul(hidden_states.permute(1, 2, 0), all_coefs_T)
    # [altup_num_inputs, num_tokens, hidden_size]
    predictions = predictions.permute(2, 0, 1)
    predictions += hidden_states
    return predictions.contiguous()

scale_corrected_output 

scale_corrected_output(corrected: Tensor) -> Tensor
Source code in vllm/model_executor/models/gemma3n.py 
114
115
116
def scale_corrected_output(self, corrected: torch.Tensor) -> torch.Tensor:
    return (corrected.type_as(self.correct_output_scale) *
            self.correct_output_scale).type_as(corrected)

Gemma3nAttention

Bases: Module

Source code in vllm/model_executor/models/gemma3n.py 
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
class Gemma3nAttention(nn.Module):

    def __init__(self,
                 config: Gemma3nTextConfig,
                 hidden_size: int,
                 num_heads: int,
                 num_kv_heads: int,
                 head_dim: int,
                 max_position_embeddings: int,
                 cache_config: Optional[CacheConfig] = None,
                 quant_config: Optional[QuantizationConfig] = None,
                 prefix: str = "") -> None:
        super().__init__()
        self.config = config
        self.hidden_size = hidden_size
        tp_size = get_tensor_model_parallel_world_size()
        self.total_num_heads = num_heads
        assert self.total_num_heads % tp_size == 0
        self.num_heads = self.total_num_heads // tp_size
        self.total_num_kv_heads = num_kv_heads
        if self.total_num_kv_heads >= tp_size:
            # Number of KV heads is greater than TP size, so we partition
            # the KV heads across multiple tensor parallel GPUs.
            assert self.total_num_kv_heads % tp_size == 0
        else:
            # Number of KV heads is less than TP size, so we replicate
            # the KV heads across multiple tensor parallel GPUs.
            assert tp_size % self.total_num_kv_heads == 0
        self.num_kv_heads = max(1, self.total_num_kv_heads // tp_size)
        self.head_dim = head_dim
        self.q_size = self.num_heads * self.head_dim
        self.kv_size = self.num_kv_heads * self.head_dim

        self.qkv_proj = QKVParallelLinear(
            hidden_size,
            self.head_dim,
            self.total_num_heads,
            self.total_num_kv_heads,
            bias=config.attention_bias,
            quant_config=quant_config,
            prefix=f"{prefix}.qkv_proj",
        )
        self.o_proj = RowParallelLinear(
            self.total_num_heads * self.head_dim,
            hidden_size,
            bias=config.attention_bias,
            quant_config=quant_config,
            prefix=f"{prefix}.o_proj",
        )
        self.q_norm = RMSNorm(hidden_size=self.head_dim,
                              eps=config.rms_norm_eps)
        self.k_norm = RMSNorm(hidden_size=self.head_dim,
                              eps=config.rms_norm_eps)
        self.v_norm = RMSNorm(hidden_size=self.head_dim,
                              eps=config.rms_norm_eps,
                              has_weight=False)

        layer_idx = extract_layer_index(prefix)
        if config.layer_types[layer_idx] == "sliding_attention":
            self.sliding_window = config.sliding_window
            rope_theta = config.rope_local_base_freq
            rope_scaling = {"rope_type": "default"}
        else:
            self.sliding_window = None
            rope_theta = config.rope_theta
            rope_scaling = config.rope_scaling

        first_kv_shared_layer_idx = (config.num_hidden_layers -
                                     config.num_kv_shared_layers)
        self.is_kv_shared = layer_idx >= first_kv_shared_layer_idx

        if self.is_kv_shared:
            # Last full attention layer is 1 before sharing
            # Last sliding attention layer is 2 before sharing
            offset = 2 if self.sliding_window is not None else 1
            kv_shared_layer_index = first_kv_shared_layer_idx - offset
            kv_sharing_target_layer_name = f"model.language_model.layers.{kv_shared_layer_index}.self_attn.attn"  # noqa: E501
        else:
            kv_sharing_target_layer_name = None

        self.rotary_emb = get_rope(
            self.head_dim,
            rotary_dim=self.head_dim,
            max_position=max_position_embeddings,
            base=rope_theta,
            is_neox_style=True,
            rope_scaling=rope_scaling,
        )

        self.attn = Attention(
            num_heads=self.num_heads,
            head_size=self.head_dim,
            scale=1.0,
            num_kv_heads=self.num_kv_heads,
            cache_config=cache_config,
            quant_config=quant_config,
            per_layer_sliding_window=self.sliding_window,
            kv_sharing_target_layer_name=kv_sharing_target_layer_name,
            prefix=f"{prefix}.attn")

    def forward(
        self,
        positions: torch.Tensor,
        hidden_states: torch.Tensor,
        **kwargs,
    ) -> torch.Tensor:
        qkv, _ = self.qkv_proj(hidden_states)
        q, k, v = qkv.split([self.q_size, self.kv_size, self.kv_size], dim=-1)

        q = q.unflatten(-1, (self.num_heads, self.head_dim))
        q = self.q_norm(q)
        q = q.flatten(-2, -1)
        k = k.unflatten(-1, (self.num_kv_heads, self.head_dim))
        k = self.k_norm(k)
        k = k.flatten(-2, -1)
        v = v.unflatten(-1, (self.num_kv_heads, self.head_dim))
        v = self.v_norm(v)
        v = v.flatten(-2, -1)

        q, k = self.rotary_emb(positions, q, k)
        attn_output = self.attn(q, k, v)

        output, _ = self.o_proj(attn_output)
        return output

attn instance-attribute 

attn = Attention(
    num_heads=num_heads,
    head_size=head_dim,
    scale=1.0,
    num_kv_heads=num_kv_heads,
    cache_config=cache_config,
    quant_config=quant_config,
    per_layer_sliding_window=sliding_window,
    kv_sharing_target_layer_name=kv_sharing_target_layer_name,
    prefix=f"{prefix}.attn",
)

config instance-attribute 

config = config

head_dim instance-attribute 

head_dim = head_dim

hidden_size instance-attribute 

hidden_size = hidden_size

is_kv_shared instance-attribute 

is_kv_shared = layer_idx >= first_kv_shared_layer_idx

k_norm instance-attribute 

k_norm = RMSNorm(hidden_size=head_dim, eps=rms_norm_eps)

kv_size instance-attribute 

kv_size = num_kv_heads * head_dim

num_heads instance-attribute 

num_heads = total_num_heads // tp_size

num_kv_heads instance-attribute 

num_kv_heads = max(1, total_num_kv_heads // tp_size)

o_proj instance-attribute 

o_proj = RowParallelLinear(
    total_num_heads * head_dim,
    hidden_size,
    bias=attention_bias,
    quant_config=quant_config,
    prefix=f"{prefix}.o_proj",
)

q_norm instance-attribute 

q_norm = RMSNorm(hidden_size=head_dim, eps=rms_norm_eps)

q_size instance-attribute 

q_size = num_heads * head_dim

qkv_proj instance-attribute 

qkv_proj = QKVParallelLinear(
    hidden_size,
    head_dim,
    total_num_heads,
    total_num_kv_heads,
    bias=attention_bias,
    quant_config=quant_config,
    prefix=f"{prefix}.qkv_proj",
)

rotary_emb instance-attribute 

rotary_emb = get_rope(
    head_dim,
    rotary_dim=head_dim,
    max_position=max_position_embeddings,
    base=rope_theta,
    is_neox_style=True,
    rope_scaling=rope_scaling,
)

sliding_window instance-attribute 

sliding_window = sliding_window

total_num_heads instance-attribute 

total_num_heads = num_heads

total_num_kv_heads instance-attribute 

total_num_kv_heads = num_kv_heads

v_norm instance-attribute 

v_norm = RMSNorm(
    hidden_size=head_dim, eps=rms_norm_eps, has_weight=False
)

__init__ 

__init__(
    config: Gemma3nTextConfig,
    hidden_size: int,
    num_heads: int,
    num_kv_heads: int,
    head_dim: int,
    max_position_embeddings: int,
    cache_config: Optional[CacheConfig] = None,
    quant_config: Optional[QuantizationConfig] = None,
    prefix: str = "",
) -> None
Source code in vllm/model_executor/models/gemma3n.py 
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
def __init__(self,
             config: Gemma3nTextConfig,
             hidden_size: int,
             num_heads: int,
             num_kv_heads: int,
             head_dim: int,
             max_position_embeddings: int,
             cache_config: Optional[CacheConfig] = None,
             quant_config: Optional[QuantizationConfig] = None,
             prefix: str = "") -> None:
    super().__init__()
    self.config = config
    self.hidden_size = hidden_size
    tp_size = get_tensor_model_parallel_world_size()
    self.total_num_heads = num_heads
    assert self.total_num_heads % tp_size == 0
    self.num_heads = self.total_num_heads // tp_size
    self.total_num_kv_heads = num_kv_heads
    if self.total_num_kv_heads >= tp_size:
        # Number of KV heads is greater than TP size, so we partition
        # the KV heads across multiple tensor parallel GPUs.
        assert self.total_num_kv_heads % tp_size == 0
    else:
        # Number of KV heads is less than TP size, so we replicate
        # the KV heads across multiple tensor parallel GPUs.
        assert tp_size % self.total_num_kv_heads == 0
    self.num_kv_heads = max(1, self.total_num_kv_heads // tp_size)
    self.head_dim = head_dim
    self.q_size = self.num_heads * self.head_dim
    self.kv_size = self.num_kv_heads * self.head_dim

    self.qkv_proj = QKVParallelLinear(
        hidden_size,
        self.head_dim,
        self.total_num_heads,
        self.total_num_kv_heads,
        bias=config.attention_bias,
        quant_config=quant_config,
        prefix=f"{prefix}.qkv_proj",
    )
    self.o_proj = RowParallelLinear(
        self.total_num_heads * self.head_dim,
        hidden_size,
        bias=config.attention_bias,
        quant_config=quant_config,
        prefix=f"{prefix}.o_proj",
    )
    self.q_norm = RMSNorm(hidden_size=self.head_dim,
                          eps=config.rms_norm_eps)
    self.k_norm = RMSNorm(hidden_size=self.head_dim,
                          eps=config.rms_norm_eps)
    self.v_norm = RMSNorm(hidden_size=self.head_dim,
                          eps=config.rms_norm_eps,
                          has_weight=False)

    layer_idx = extract_layer_index(prefix)
    if config.layer_types[layer_idx] == "sliding_attention":
        self.sliding_window = config.sliding_window
        rope_theta = config.rope_local_base_freq
        rope_scaling = {"rope_type": "default"}
    else:
        self.sliding_window = None
        rope_theta = config.rope_theta
        rope_scaling = config.rope_scaling

    first_kv_shared_layer_idx = (config.num_hidden_layers -
                                 config.num_kv_shared_layers)
    self.is_kv_shared = layer_idx >= first_kv_shared_layer_idx

    if self.is_kv_shared:
        # Last full attention layer is 1 before sharing
        # Last sliding attention layer is 2 before sharing
        offset = 2 if self.sliding_window is not None else 1
        kv_shared_layer_index = first_kv_shared_layer_idx - offset
        kv_sharing_target_layer_name = f"model.language_model.layers.{kv_shared_layer_index}.self_attn.attn"  # noqa: E501
    else:
        kv_sharing_target_layer_name = None

    self.rotary_emb = get_rope(
        self.head_dim,
        rotary_dim=self.head_dim,
        max_position=max_position_embeddings,
        base=rope_theta,
        is_neox_style=True,
        rope_scaling=rope_scaling,
    )

    self.attn = Attention(
        num_heads=self.num_heads,
        head_size=self.head_dim,
        scale=1.0,
        num_kv_heads=self.num_kv_heads,
        cache_config=cache_config,
        quant_config=quant_config,
        per_layer_sliding_window=self.sliding_window,
        kv_sharing_target_layer_name=kv_sharing_target_layer_name,
        prefix=f"{prefix}.attn")

forward 

forward(
    positions: Tensor, hidden_states: Tensor, **kwargs
) -> Tensor
Source code in vllm/model_executor/models/gemma3n.py 
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
def forward(
    self,
    positions: torch.Tensor,
    hidden_states: torch.Tensor,
    **kwargs,
) -> torch.Tensor:
    qkv, _ = self.qkv_proj(hidden_states)
    q, k, v = qkv.split([self.q_size, self.kv_size, self.kv_size], dim=-1)

    q = q.unflatten(-1, (self.num_heads, self.head_dim))
    q = self.q_norm(q)
    q = q.flatten(-2, -1)
    k = k.unflatten(-1, (self.num_kv_heads, self.head_dim))
    k = self.k_norm(k)
    k = k.flatten(-2, -1)
    v = v.unflatten(-1, (self.num_kv_heads, self.head_dim))
    v = self.v_norm(v)
    v = v.flatten(-2, -1)

    q, k = self.rotary_emb(positions, q, k)
    attn_output = self.attn(q, k, v)

    output, _ = self.o_proj(attn_output)
    return output

Gemma3nDecoderLayer

Bases: Module

Source code in vllm/model_executor/models/gemma3n.py 
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
class Gemma3nDecoderLayer(nn.Module):

    def __init__(
        self,
        config: Gemma3nTextConfig,
        cache_config: Optional[CacheConfig] = None,
        quant_config: Optional[QuantizationConfig] = None,
        prefix: str = "",
    ) -> None:
        super().__init__()
        self.altup_active_idx = config.altup_active_idx
        assert config.altup_correct_scale

        self.altup = Gemma3nAltUp(
            hidden_size=config.hidden_size,
            rms_norm_eps=config.rms_norm_eps,
            altup_num_inputs=config.altup_num_inputs,
            altup_coef_clip=config.altup_coef_clip,
            altup_active_idx=config.altup_active_idx,
            prefix=f"{prefix}.altup",
        )
        self.self_attn = Gemma3nAttention(
            config=config,
            hidden_size=config.hidden_size,
            num_heads=config.num_attention_heads,
            num_kv_heads=config.num_key_value_heads,
            head_dim=config.head_dim,
            max_position_embeddings=config.max_position_embeddings,
            cache_config=cache_config,
            quant_config=quant_config,
            prefix=f"{prefix}.self_attn",
        )
        self.mlp = Gemma3nMLP(
            hidden_size=config.hidden_size,
            # NOTE: Matformer https://github.com/huggingface/transformers/blob/a52478253bbe522a420e88ea3940d4d98a935300/src/transformers/models/gemma3n/modular_gemma3n.py#L258 # noqa: E501
            intermediate_size=config.intermediate_size[extract_layer_index(
                prefix)],
            hidden_activation=config.hidden_activation,
            quant_config=quant_config,
            activation_sparsity=config.activation_sparsity_pattern[
                extract_layer_index(prefix)],
            prefix=f"{prefix}.mlp",
        )
        self.laurel = Gemma3nLaurelBlock(
            hidden_size=config.hidden_size,
            laurel_rank=config.laurel_rank,
            rms_norm_eps=config.rms_norm_eps,
            prefix=f"{prefix}.laurel",
        )

        # NOTE(rob): should be ColumnParallelLinear and RowParallelLinear
        # But, we need to add per_layer_input_gate(x) to per_layer_input.
        # per_layer_input cannot be sharded, so we replicate for now.
        self.per_layer_input_gate = ReplicatedLinear(
            config.hidden_size,
            config.hidden_size_per_layer_input,
            bias=False,
            prefix=f"{prefix}.per_layer_input_gate",
            return_bias=False,
        )
        self.per_layer_projection = ReplicatedLinear(
            config.hidden_size_per_layer_input,
            config.hidden_size,
            bias=False,
            prefix=f"{prefix}.per_layer_projection",
            return_bias=False,
        )

        # LayerNorms.
        self.input_layernorm = RMSNorm(
            config.hidden_size,
            eps=config.rms_norm_eps,
        )
        self.post_attention_layernorm = RMSNorm(
            config.hidden_size,
            eps=config.rms_norm_eps,
        )
        self.pre_feedforward_layernorm = RMSNorm(
            config.hidden_size,
            eps=config.rms_norm_eps,
        )
        self.post_feedforward_layernorm = RMSNorm(
            config.hidden_size,
            eps=config.rms_norm_eps,
        )
        self.post_per_layer_input_norm = RMSNorm(
            config.hidden_size,
            eps=config.rms_norm_eps,
        )

        self.act_fn = _ACTIVATION_REGISTRY[config.hidden_activation]

    def forward(
        self,
        positions: torch.Tensor,
        hidden_states: torch.Tensor,
        per_layer_input: torch.Tensor,
        **kwargs,
    ) -> tuple[torch.Tensor, torch.Tensor]:

        # ActUp (predict).
        predictions = self.altup.predict(hidden_states)
        active_prediction = predictions[self.altup_active_idx]
        active_prediction_normed = self.input_layernorm(active_prediction)
        laurel_output = self.laurel(active_prediction_normed)

        # Attention.
        attn = self.self_attn(
            positions=positions,
            hidden_states=active_prediction_normed,
            **kwargs,
        )
        attn = self.post_attention_layernorm(attn)
        attn_gated = attn + active_prediction
        attn_laurel = (attn_gated + laurel_output) / torch.sqrt(
            torch.tensor(2.0))

        # MLP.
        attn_norm = self.pre_feedforward_layernorm(attn_laurel)
        attn_ffw = self.mlp(attn_norm)
        attn_ffw_norm = self.post_feedforward_layernorm(attn_ffw)
        attn_ffw_laurel_gated = attn_laurel + attn_ffw_norm

        # ActUp (connect).
        corrected_predictions = self.altup.correct(predictions,
                                                   attn_ffw_laurel_gated)
        first_prediction = corrected_predictions[self.altup_active_idx]
        first_prediction = self.altup.scale_corrected_output(first_prediction)

        # per_layer_input_gate adapted from jax.numpy.einsum("btd,dp->btp", ...)
        first_prediction = self.per_layer_input_gate(first_prediction)
        first_prediction = self.act_fn(first_prediction)
        first_prediction = torch.mul(first_prediction, per_layer_input)

        # per_layer_projection adapted from jax.numpy.einsum("btp,pd->btd", ...)
        first_prediction = self.per_layer_projection(first_prediction)
        first_prediction = self.post_per_layer_input_norm(first_prediction)
        corrected_predictions[1:] += first_prediction

        return corrected_predictions

act_fn instance-attribute 

act_fn = _ACTIVATION_REGISTRY[hidden_activation]

altup instance-attribute 

altup = Gemma3nAltUp(
    hidden_size=hidden_size,
    rms_norm_eps=rms_norm_eps,
    altup_num_inputs=altup_num_inputs,
    altup_coef_clip=altup_coef_clip,
    altup_active_idx=altup_active_idx,
    prefix=f"{prefix}.altup",
)

altup_active_idx instance-attribute 

altup_active_idx = altup_active_idx

input_layernorm instance-attribute 

input_layernorm = RMSNorm(hidden_size, eps=rms_norm_eps)

laurel instance-attribute 

laurel = Gemma3nLaurelBlock(
    hidden_size=hidden_size,
    laurel_rank=laurel_rank,
    rms_norm_eps=rms_norm_eps,
    prefix=f"{prefix}.laurel",
)

mlp instance-attribute 

mlp = Gemma3nMLP(
    hidden_size=hidden_size,
    intermediate_size=intermediate_size[
        extract_layer_index(prefix)
    ],
    hidden_activation=hidden_activation,
    quant_config=quant_config,
    activation_sparsity=activation_sparsity_pattern[
        extract_layer_index(prefix)
    ],
    prefix=f"{prefix}.mlp",
)

per_layer_input_gate instance-attribute 

per_layer_input_gate = ReplicatedLinear(
    hidden_size,
    hidden_size_per_layer_input,
    bias=False,
    prefix=f"{prefix}.per_layer_input_gate",
    return_bias=False,
)

per_layer_projection instance-attribute 

per_layer_projection = ReplicatedLinear(
    hidden_size_per_layer_input,
    hidden_size,
    bias=False,
    prefix=f"{prefix}.per_layer_projection",
    return_bias=False,
)

post_attention_layernorm instance-attribute 

post_attention_layernorm = RMSNorm(
    hidden_size, eps=rms_norm_eps
)

post_feedforward_layernorm instance-attribute 

post_feedforward_layernorm = RMSNorm(
    hidden_size, eps=rms_norm_eps
)

post_per_layer_input_norm instance-attribute 

post_per_layer_input_norm = RMSNorm(
    hidden_size, eps=rms_norm_eps
)

pre_feedforward_layernorm instance-attribute 

pre_feedforward_layernorm = RMSNorm(
    hidden_size, eps=rms_norm_eps
)

self_attn instance-attribute 

self_attn = Gemma3nAttention(
    config=config,
    hidden_size=hidden_size,
    num_heads=num_attention_heads,
    num_kv_heads=num_key_value_heads,
    head_dim=head_dim,
    max_position_embeddings=max_position_embeddings,
    cache_config=cache_config,
    quant_config=quant_config,
    prefix=f"{prefix}.self_attn",
)

__init__ 

__init__(
    config: Gemma3nTextConfig,
    cache_config: Optional[CacheConfig] = None,
    quant_config: Optional[QuantizationConfig] = None,
    prefix: str = "",
) -> None
Source code in vllm/model_executor/models/gemma3n.py 
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
def __init__(
    self,
    config: Gemma3nTextConfig,
    cache_config: Optional[CacheConfig] = None,
    quant_config: Optional[QuantizationConfig] = None,
    prefix: str = "",
) -> None:
    super().__init__()
    self.altup_active_idx = config.altup_active_idx
    assert config.altup_correct_scale

    self.altup = Gemma3nAltUp(
        hidden_size=config.hidden_size,
        rms_norm_eps=config.rms_norm_eps,
        altup_num_inputs=config.altup_num_inputs,
        altup_coef_clip=config.altup_coef_clip,
        altup_active_idx=config.altup_active_idx,
        prefix=f"{prefix}.altup",
    )
    self.self_attn = Gemma3nAttention(
        config=config,
        hidden_size=config.hidden_size,
        num_heads=config.num_attention_heads,
        num_kv_heads=config.num_key_value_heads,
        head_dim=config.head_dim,
        max_position_embeddings=config.max_position_embeddings,
        cache_config=cache_config,
        quant_config=quant_config,
        prefix=f"{prefix}.self_attn",
    )
    self.mlp = Gemma3nMLP(
        hidden_size=config.hidden_size,
        # NOTE: Matformer https://github.com/huggingface/transformers/blob/a52478253bbe522a420e88ea3940d4d98a935300/src/transformers/models/gemma3n/modular_gemma3n.py#L258 # noqa: E501
        intermediate_size=config.intermediate_size[extract_layer_index(
            prefix)],
        hidden_activation=config.hidden_activation,
        quant_config=quant_config,
        activation_sparsity=config.activation_sparsity_pattern[
            extract_layer_index(prefix)],
        prefix=f"{prefix}.mlp",
    )
    self.laurel = Gemma3nLaurelBlock(
        hidden_size=config.hidden_size,
        laurel_rank=config.laurel_rank,
        rms_norm_eps=config.rms_norm_eps,
        prefix=f"{prefix}.laurel",
    )

    # NOTE(rob): should be ColumnParallelLinear and RowParallelLinear
    # But, we need to add per_layer_input_gate(x) to per_layer_input.
    # per_layer_input cannot be sharded, so we replicate for now.
    self.per_layer_input_gate = ReplicatedLinear(
        config.hidden_size,
        config.hidden_size_per_layer_input,
        bias=False,
        prefix=f"{prefix}.per_layer_input_gate",
        return_bias=False,
    )
    self.per_layer_projection = ReplicatedLinear(
        config.hidden_size_per_layer_input,
        config.hidden_size,
        bias=False,
        prefix=f"{prefix}.per_layer_projection",
        return_bias=False,
    )

    # LayerNorms.
    self.input_layernorm = RMSNorm(
        config.hidden_size,
        eps=config.rms_norm_eps,
    )
    self.post_attention_layernorm = RMSNorm(
        config.hidden_size,
        eps=config.rms_norm_eps,
    )
    self.pre_feedforward_layernorm = RMSNorm(
        config.hidden_size,
        eps=config.rms_norm_eps,
    )
    self.post_feedforward_layernorm = RMSNorm(
        config.hidden_size,
        eps=config.rms_norm_eps,
    )
    self.post_per_layer_input_norm = RMSNorm(
        config.hidden_size,
        eps=config.rms_norm_eps,
    )

    self.act_fn = _ACTIVATION_REGISTRY[config.hidden_activation]

forward 

forward(
    positions: Tensor,
    hidden_states: Tensor,
    per_layer_input: Tensor,
    **kwargs,
) -> tuple[Tensor, Tensor]
Source code in vllm/model_executor/models/gemma3n.py 
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
def forward(
    self,
    positions: torch.Tensor,
    hidden_states: torch.Tensor,
    per_layer_input: torch.Tensor,
    **kwargs,
) -> tuple[torch.Tensor, torch.Tensor]:

    # ActUp (predict).
    predictions = self.altup.predict(hidden_states)
    active_prediction = predictions[self.altup_active_idx]
    active_prediction_normed = self.input_layernorm(active_prediction)
    laurel_output = self.laurel(active_prediction_normed)

    # Attention.
    attn = self.self_attn(
        positions=positions,
        hidden_states=active_prediction_normed,
        **kwargs,
    )
    attn = self.post_attention_layernorm(attn)
    attn_gated = attn + active_prediction
    attn_laurel = (attn_gated + laurel_output) / torch.sqrt(
        torch.tensor(2.0))

    # MLP.
    attn_norm = self.pre_feedforward_layernorm(attn_laurel)
    attn_ffw = self.mlp(attn_norm)
    attn_ffw_norm = self.post_feedforward_layernorm(attn_ffw)
    attn_ffw_laurel_gated = attn_laurel + attn_ffw_norm

    # ActUp (connect).
    corrected_predictions = self.altup.correct(predictions,
                                               attn_ffw_laurel_gated)
    first_prediction = corrected_predictions[self.altup_active_idx]
    first_prediction = self.altup.scale_corrected_output(first_prediction)

    # per_layer_input_gate adapted from jax.numpy.einsum("btd,dp->btp", ...)
    first_prediction = self.per_layer_input_gate(first_prediction)
    first_prediction = self.act_fn(first_prediction)
    first_prediction = torch.mul(first_prediction, per_layer_input)

    # per_layer_projection adapted from jax.numpy.einsum("btp,pd->btd", ...)
    first_prediction = self.per_layer_projection(first_prediction)
    first_prediction = self.post_per_layer_input_norm(first_prediction)
    corrected_predictions[1:] += first_prediction

    return corrected_predictions

Gemma3nForConditionalGeneration

Bases: Module

Source code in vllm/model_executor/models/gemma3n.py 
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
class Gemma3nForConditionalGeneration(nn.Module):
    packed_modules_mapping = {
        "qkv_proj": [
            "q_proj",
            "k_proj",
            "v_proj",
        ],
        "gate_up_proj": [
            "gate_proj",
            "up_proj",
        ],
    }

    def __init__(self, *, vllm_config: VllmConfig, prefix: str = ""):
        config = vllm_config.model_config.hf_config
        lora_config = vllm_config.lora_config
        del lora_config  # Unused.
        super().__init__()
        self.config = config
        self.model = Gemma3nModel(vllm_config=vllm_config,
                                  prefix=maybe_prefix(prefix, "model"))
        self.logits_processor = LogitsProcessor(
            config.text_config.vocab_size,
            soft_cap=config.text_config.final_logit_softcapping)

    def get_input_embeddings(self, input_ids: torch.Tensor) -> torch.Tensor:
        return self.model.language_model.get_input_embeddings(input_ids)

    def forward(
        self,
        input_ids: torch.Tensor,
        positions: torch.Tensor,
        intermediate_tensors: Optional[IntermediateTensors] = None,
        inputs_embeds: Optional[torch.Tensor] = None,
        **kwargs,
    ) -> Union[torch.Tensor, IntermediateTensors]:
        hidden_states = self.model(input_ids, positions, intermediate_tensors,
                                   inputs_embeds, **kwargs)
        return hidden_states

    def compute_logits(
        self,
        hidden_states: torch.Tensor,
        sampling_metadata: Optional[SamplingMetadata],
    ) -> Optional[torch.Tensor]:
        logits = self.logits_processor(self.model.language_model.embed_tokens,
                                       hidden_states, sampling_metadata)
        return logits

    def load_weights(self, weights: Iterable[tuple[str,
                                                   torch.Tensor]]) -> set[str]:
        loader = AutoWeightsLoader(self,
                                   skip_substrs=([
                                       "embed_audio.", "embed_vision.",
                                       "audio_tower.", "vision_tower."
                                   ]))
        return loader.load_weights(weights)

config instance-attribute 

config = config

logits_processor instance-attribute 

logits_processor = LogitsProcessor(
    vocab_size, soft_cap=final_logit_softcapping
)

model instance-attribute 

model = Gemma3nModel(
    vllm_config=vllm_config,
    prefix=maybe_prefix(prefix, "model"),
)

packed_modules_mapping class-attribute instance-attribute 

packed_modules_mapping = {
    "qkv_proj": ["q_proj", "k_proj", "v_proj"],
    "gate_up_proj": ["gate_proj", "up_proj"],
}

__init__ 

__init__(*, vllm_config: VllmConfig, prefix: str = '')
Source code in vllm/model_executor/models/gemma3n.py 
768
769
770
771
772
773
774
775
776
777
778
def __init__(self, *, vllm_config: VllmConfig, prefix: str = ""):
    config = vllm_config.model_config.hf_config
    lora_config = vllm_config.lora_config
    del lora_config  # Unused.
    super().__init__()
    self.config = config
    self.model = Gemma3nModel(vllm_config=vllm_config,
                              prefix=maybe_prefix(prefix, "model"))
    self.logits_processor = LogitsProcessor(
        config.text_config.vocab_size,
        soft_cap=config.text_config.final_logit_softcapping)

compute_logits 

compute_logits(
    hidden_states: Tensor,
    sampling_metadata: Optional[SamplingMetadata],
) -> Optional[Tensor]
Source code in vllm/model_executor/models/gemma3n.py 
795
796
797
798
799
800
801
802
def compute_logits(
    self,
    hidden_states: torch.Tensor,
    sampling_metadata: Optional[SamplingMetadata],
) -> Optional[torch.Tensor]:
    logits = self.logits_processor(self.model.language_model.embed_tokens,
                                   hidden_states, sampling_metadata)
    return logits

forward 

forward(
    input_ids: Tensor,
    positions: Tensor,
    intermediate_tensors: Optional[
        IntermediateTensors
    ] = None,
    inputs_embeds: Optional[Tensor] = None,
    **kwargs,
) -> Union[Tensor, IntermediateTensors]
Source code in vllm/model_executor/models/gemma3n.py 
783
784
785
786
787
788
789
790
791
792
793
def forward(
    self,
    input_ids: torch.Tensor,
    positions: torch.Tensor,
    intermediate_tensors: Optional[IntermediateTensors] = None,
    inputs_embeds: Optional[torch.Tensor] = None,
    **kwargs,
) -> Union[torch.Tensor, IntermediateTensors]:
    hidden_states = self.model(input_ids, positions, intermediate_tensors,
                               inputs_embeds, **kwargs)
    return hidden_states

get_input_embeddings 

get_input_embeddings(input_ids: Tensor) -> Tensor
Source code in vllm/model_executor/models/gemma3n.py 
780
781
def get_input_embeddings(self, input_ids: torch.Tensor) -> torch.Tensor:
    return self.model.language_model.get_input_embeddings(input_ids)

load_weights 

load_weights(
    weights: Iterable[tuple[str, Tensor]],
) -> set[str]
Source code in vllm/model_executor/models/gemma3n.py 
804
805
806
807
808
809
810
811
def load_weights(self, weights: Iterable[tuple[str,
                                               torch.Tensor]]) -> set[str]:
    loader = AutoWeightsLoader(self,
                               skip_substrs=([
                                   "embed_audio.", "embed_vision.",
                                   "audio_tower.", "vision_tower."
                               ]))
    return loader.load_weights(weights)

Gemma3nLaurelBlock

Bases: Module

Learned Augmented Residual Layer

Source code in vllm/model_executor/models/gemma3n.py 
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
class Gemma3nLaurelBlock(nn.Module):
    """Learned Augmented Residual Layer"""

    def __init__(self, hidden_size: int, laurel_rank: int, rms_norm_eps: float,
                 prefix: str):
        super().__init__()

        self.linear_left = ColumnParallelLinear(
            hidden_size,
            laurel_rank,
            bias=False,
            prefix=f"{prefix}.linear_left",
            return_bias=False,
        )
        self.linear_right = RowParallelLinear(laurel_rank,
                                              hidden_size,
                                              bias=False,
                                              prefix=f"{prefix}.linear_right",
                                              return_bias=False)
        self.post_laurel_norm = RMSNorm(
            hidden_size=hidden_size,
            eps=rms_norm_eps,
        )

    def forward(self, x: torch.Tensor) -> torch.Tensor:
        laurel_x = self.linear_left(x)
        laurel_x = self.linear_right(laurel_x)
        normed_laurel_x = self.post_laurel_norm(laurel_x)
        return x + normed_laurel_x

linear_left instance-attribute 

linear_left = ColumnParallelLinear(
    hidden_size,
    laurel_rank,
    bias=False,
    prefix=f"{prefix}.linear_left",
    return_bias=False,
)

linear_right instance-attribute 

linear_right = RowParallelLinear(
    laurel_rank,
    hidden_size,
    bias=False,
    prefix=f"{prefix}.linear_right",
    return_bias=False,
)

post_laurel_norm instance-attribute 

post_laurel_norm = RMSNorm(
    hidden_size=hidden_size, eps=rms_norm_eps
)

__init__ 

__init__(
    hidden_size: int,
    laurel_rank: int,
    rms_norm_eps: float,
    prefix: str,
)
Source code in vllm/model_executor/models/gemma3n.py 
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
def __init__(self, hidden_size: int, laurel_rank: int, rms_norm_eps: float,
             prefix: str):
    super().__init__()

    self.linear_left = ColumnParallelLinear(
        hidden_size,
        laurel_rank,
        bias=False,
        prefix=f"{prefix}.linear_left",
        return_bias=False,
    )
    self.linear_right = RowParallelLinear(laurel_rank,
                                          hidden_size,
                                          bias=False,
                                          prefix=f"{prefix}.linear_right",
                                          return_bias=False)
    self.post_laurel_norm = RMSNorm(
        hidden_size=hidden_size,
        eps=rms_norm_eps,
    )

forward 

forward(x: Tensor) -> Tensor
Source code in vllm/model_executor/models/gemma3n.py 
191
192
193
194
195
def forward(self, x: torch.Tensor) -> torch.Tensor:
    laurel_x = self.linear_left(x)
    laurel_x = self.linear_right(laurel_x)
    normed_laurel_x = self.post_laurel_norm(laurel_x)
    return x + normed_laurel_x

Gemma3nMLP

Bases: Module

Source code in vllm/model_executor/models/gemma3n.py 
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
class Gemma3nMLP(nn.Module):

    def __init__(
        self,
        hidden_size: int,
        intermediate_size: int,
        hidden_activation: str,
        activation_sparsity: float = 0.0,
        quant_config: Optional[QuantizationConfig] = None,
        prefix: str = "",
    ) -> None:
        super().__init__()
        self.gate_up_proj = MergedColumnParallelLinear(
            hidden_size,
            [intermediate_size] * 2,
            bias=False,
            quant_config=quant_config,
            prefix=f"{prefix}.gate_up_proj",
        )
        self.down_proj = RowParallelLinear(
            intermediate_size,
            hidden_size,
            bias=False,
            quant_config=quant_config,
            prefix=f"{prefix}.down_proj",
        )
        if hidden_activation != "gelu_pytorch_tanh":
            raise ValueError(
                "Gemma3 uses `gelu_pytorch_tanh` as the hidden activation "
                "function. Please set `hidden_act` and `hidden_activation` to "
                "`gelu_pytorch_tanh`.")

        self.act_fn = GeluAndMulSparse(
            activation_sparsity=activation_sparsity,
            approximate="tanh") if activation_sparsity > 0.0 else GeluAndMul(
                approximate="tanh")

    def forward(self, x: torch.Tensor) -> torch.Tensor:
        gate_up, _ = self.gate_up_proj(x)
        x = self.act_fn(gate_up)
        x, _ = self.down_proj(x)
        return x

act_fn instance-attribute 

act_fn = (
    GeluAndMulSparse(
        activation_sparsity=activation_sparsity,
        approximate="tanh",
    )
    if activation_sparsity > 0.0
    else GeluAndMul(approximate="tanh")
)

down_proj instance-attribute 

down_proj = RowParallelLinear(
    intermediate_size,
    hidden_size,
    bias=False,
    quant_config=quant_config,
    prefix=f"{prefix}.down_proj",
)

gate_up_proj instance-attribute 

gate_up_proj = MergedColumnParallelLinear(
    hidden_size,
    [intermediate_size] * 2,
    bias=False,
    quant_config=quant_config,
    prefix=f"{prefix}.gate_up_proj",
)

__init__ 

__init__(
    hidden_size: int,
    intermediate_size: int,
    hidden_activation: str,
    activation_sparsity: float = 0.0,
    quant_config: Optional[QuantizationConfig] = None,
    prefix: str = "",
) -> None
Source code in vllm/model_executor/models/gemma3n.py 
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
def __init__(
    self,
    hidden_size: int,
    intermediate_size: int,
    hidden_activation: str,
    activation_sparsity: float = 0.0,
    quant_config: Optional[QuantizationConfig] = None,
    prefix: str = "",
) -> None:
    super().__init__()
    self.gate_up_proj = MergedColumnParallelLinear(
        hidden_size,
        [intermediate_size] * 2,
        bias=False,
        quant_config=quant_config,
        prefix=f"{prefix}.gate_up_proj",
    )
    self.down_proj = RowParallelLinear(
        intermediate_size,
        hidden_size,
        bias=False,
        quant_config=quant_config,
        prefix=f"{prefix}.down_proj",
    )
    if hidden_activation != "gelu_pytorch_tanh":
        raise ValueError(
            "Gemma3 uses `gelu_pytorch_tanh` as the hidden activation "
            "function. Please set `hidden_act` and `hidden_activation` to "
            "`gelu_pytorch_tanh`.")

    self.act_fn = GeluAndMulSparse(
        activation_sparsity=activation_sparsity,
        approximate="tanh") if activation_sparsity > 0.0 else GeluAndMul(
            approximate="tanh")

forward 

forward(x: Tensor) -> Tensor
Source code in vllm/model_executor/models/gemma3n.py 
235
236
237
238
239
def forward(self, x: torch.Tensor) -> torch.Tensor:
    gate_up, _ = self.gate_up_proj(x)
    x = self.act_fn(gate_up)
    x, _ = self.down_proj(x)
    return x

Gemma3nModel

Bases: Module

Source code in vllm/model_executor/models/gemma3n.py 
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
class Gemma3nModel(nn.Module):

    def __init__(self, vllm_config: VllmConfig, prefix: str = ""):
        super().__init__()
        self.language_model = Gemma3nTextModel(vllm_config=vllm_config,
                                               prefix=maybe_prefix(
                                                   prefix, "language_model"))

    def forward(
        self,
        input_ids: Optional[torch.Tensor],
        positions: torch.Tensor,
        intermediate_tensors: Optional[IntermediateTensors] = None,
        inputs_embeds: Optional[torch.Tensor] = None,
        **kwargs,
    ) -> torch.Tensor:
        return self.language_model(input_ids=input_ids,
                                   positions=positions,
                                   inputs_embeds=inputs_embeds,
                                   **kwargs)

language_model instance-attribute 

language_model = Gemma3nTextModel(
    vllm_config=vllm_config,
    prefix=maybe_prefix(prefix, "language_model"),
)

__init__ 

__init__(vllm_config: VllmConfig, prefix: str = '')
Source code in vllm/model_executor/models/gemma3n.py 
735
736
737
738
739
def __init__(self, vllm_config: VllmConfig, prefix: str = ""):
    super().__init__()
    self.language_model = Gemma3nTextModel(vllm_config=vllm_config,
                                           prefix=maybe_prefix(
                                               prefix, "language_model"))

forward 

forward(
    input_ids: Optional[Tensor],
    positions: Tensor,
    intermediate_tensors: Optional[
        IntermediateTensors
    ] = None,
    inputs_embeds: Optional[Tensor] = None,
    **kwargs,
) -> Tensor
Source code in vllm/model_executor/models/gemma3n.py 
741
742
743
744
745
746
747
748
749
750
751
752
def forward(
    self,
    input_ids: Optional[torch.Tensor],
    positions: torch.Tensor,
    intermediate_tensors: Optional[IntermediateTensors] = None,
    inputs_embeds: Optional[torch.Tensor] = None,
    **kwargs,
) -> torch.Tensor:
    return self.language_model(input_ids=input_ids,
                               positions=positions,
                               inputs_embeds=inputs_embeds,
                               **kwargs)

Gemma3nTextModel

Bases: Module

Source code in vllm/model_executor/models/gemma3n.py 
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
@support_torch_compile
class Gemma3nTextModel(nn.Module):

    def __init__(self, *, vllm_config: VllmConfig, prefix: str = ""):
        super().__init__()
        config = vllm_config.model_config.hf_config.text_config
        cache_config = vllm_config.cache_config
        quant_config = vllm_config.quant_config
        self.config = config
        self.quant_config = quant_config

        self.embed_tokens = VocabParallelEmbedding(
            config.vocab_size,
            config.hidden_size,
            prefix=f"{prefix}.embed_tokens",
        )
        self.embed_scale = torch.tensor(
            config.hidden_size**0.5,
            dtype=self.embed_tokens.weight.dtype,
        )
        self.embed_tokens_per_layer = VocabParallelEmbedding(
            config.vocab_size_per_layer_input,
            config.num_hidden_layers * config.hidden_size_per_layer_input,
            prefix=f"{prefix}.per_layer_embed_tokens",
        )
        self.embed_scale_per_layer = torch.tensor(
            config.hidden_size_per_layer_input**0.5,
            dtype=self.embed_tokens.weight.dtype,
        )
        self.per_layer_model_projection = ColumnParallelLinear(
            config.hidden_size,
            config.num_hidden_layers * config.hidden_size_per_layer_input,
            bias=False,
            gather_output=True,
            return_bias=False,
            prefix=f"{prefix}.per_layer_model_projection",
        )
        self.per_layer_projection_norm = RMSNorm(
            hidden_size=config.hidden_size_per_layer_input,
            eps=config.rms_norm_eps,
        )
        self.per_layer_input_scale = torch.rsqrt(torch.tensor(2.0)).to(
            self.embed_tokens.weight.dtype)
        self.per_layer_projection_scale = torch.tensor(
            config.hidden_size**0.5,
            dtype=self.embed_tokens.weight.dtype,
        )
        self.altup_projections = nn.ModuleList([
            ColumnParallelLinear(
                config.hidden_size,
                config.hidden_size,
                bias=False,
                gather_output=True,
                return_bias=False,
                prefix=f"{prefix}.{idx-1}.altup_projections",
            ) for idx in range(1, self.config.altup_num_inputs)
        ])
        self.altup_unembed_projections = nn.ModuleList([
            ColumnParallelLinear(
                config.hidden_size,
                config.hidden_size,
                bias=False,
                gather_output=True,
                return_bias=False,
                prefix=f"{prefix}.{idx-1}.altup_unembed_projections",
            ) for idx in range(1, self.config.altup_num_inputs)
        ])

        # Transformer blocks.
        self.start_layer, self.end_layer, self.layers = make_layers(
            config.num_hidden_layers,
            lambda prefix: Gemma3nDecoderLayer(
                config, cache_config, quant_config, prefix=prefix),
            prefix=f"{prefix}.layers")
        self.norm = RMSNorm(
            config.hidden_size,
            eps=config.rms_norm_eps,
        )
        self.eps = torch.tensor(torch.finfo().min)

    def get_input_embeddings(self, input_ids: torch.Tensor) -> torch.Tensor:
        return self.embed_tokens(input_ids) * self.embed_scale

    def get_per_layer_input_embeddings(
            self, input_ids: torch.Tensor) -> torch.Tensor:
        # Deal with the fact that vocab_size_per_layer_input < vocab_size
        # which causes us to have some out of vocab tokens by setting
        # those token ids to 0. This matches the HF implementation.
        per_layer_inputs_mask = torch.logical_and(
            input_ids >= 0, input_ids < self.config.vocab_size_per_layer_input)
        per_layer_inputs_tokens = torch.where(per_layer_inputs_mask, input_ids,
                                              torch.zeros_like(input_ids))
        return self.embed_tokens_per_layer(
            per_layer_inputs_tokens) * self.embed_scale_per_layer

    def forward(
        self,
        input_ids: Optional[torch.Tensor],
        positions: torch.Tensor,
        inputs_embeds: Optional[torch.Tensor] = None,
        **kwargs,
    ) -> Union[torch.Tensor, IntermediateTensors]:
        if inputs_embeds is not None:
            hidden_states_0 = inputs_embeds
        else:
            hidden_states_0 = self.get_input_embeddings(input_ids)

        # Per layer inputs.
        if input_ids is None:
            raise ValueError("Passing None for input ids is not supported.")
        per_layer_inputs = self.get_per_layer_input_embeddings(input_ids)
        per_layer_inputs = per_layer_inputs.reshape(
            -1, self.config.num_hidden_layers,
            self.config.hidden_size_per_layer_input)
        per_layer_projection = self.per_layer_model_projection(hidden_states_0)
        per_layer_projection = per_layer_projection.reshape(
            *hidden_states_0.shape[:-1],
            self.config.num_hidden_layers,
            self.config.hidden_size_per_layer_input,
        )
        per_layer_projection = self.per_layer_projection_norm(
            per_layer_projection)
        per_layer_inputs = per_layer_projection + per_layer_inputs
        per_layer_inputs *= self.per_layer_input_scale

        # Altup embed.
        hidden_states = [hidden_states_0] * self.config.altup_num_inputs
        target_magnitude = torch.mean(hidden_states_0**2, dim=-1,
                                      keepdim=True)**0.5
        for i in range(1, self.config.altup_num_inputs):
            hidden_states[i] = self.altup_projections[i - 1](hidden_states[i])
            new_magnitude = torch.mean(hidden_states[i]**2,
                                       dim=-1,
                                       keepdim=True)**0.5
            hidden_states[i] *= target_magnitude / torch.maximum(
                new_magnitude, self.eps)
        hidden_states = torch.stack(hidden_states, dim=0)

        # Transformer blocks.
        for layer_idx, layer in enumerate(self.layers):
            # [altup_num_inputs, num_tokens, hidden_size]
            hidden_states = layer(
                positions=positions,
                hidden_states=hidden_states,
                per_layer_input=per_layer_inputs[:, layer_idx, :],
                **kwargs,
            )

        # Altup unembed.
        target_magnitude = torch.mean(hidden_states[0]**2,
                                      dim=-1,
                                      keepdim=True)**0.5
        for i in range(1, self.config.altup_num_inputs):
            hidden_states[i] = self.altup_unembed_projections[i - 1](
                hidden_states[i])
            new_magnitude = torch.mean(hidden_states[i]**2,
                                       dim=-1,
                                       keepdim=True)**0.5
            hidden_states[i] *= target_magnitude / torch.maximum(
                new_magnitude, self.eps)
        # [altup_num_inputs,num_tokens,hidden_size] -> [num_tokens,hidden_size]
        hidden_states = torch.mean(hidden_states, dim=0)

        return self.norm(hidden_states)

    def load_weights(self, weights: Iterable[tuple[str,
                                                   torch.Tensor]]) -> set[str]:
        stacked_params_mapping = [
            # (param_name, shard_name, shard_id)
            ("qkv_proj", "q_proj", "q"),
            ("qkv_proj", "k_proj", "k"),
            ("qkv_proj", "v_proj", "v"),
            ("gate_up_proj", "gate_proj", 0),
            ("gate_up_proj", "up_proj", 1),
        ]
        params_dict = dict(self.named_parameters())
        loaded_params: set[str] = set()
        for name, loaded_weight in weights:
            if (self.quant_config is not None and
                (scale_name := self.quant_config.get_cache_scale(name))):
                # Loading kv cache scales for compressed-tensors quantization
                param = params_dict[scale_name]
                weight_loader = getattr(param, "weight_loader",
                                        default_weight_loader)
                loaded_weight = loaded_weight[0]
                weight_loader(param, loaded_weight)
                loaded_params.add(scale_name)
                continue
            for (param_name, shard_name, shard_id) in stacked_params_mapping:
                if shard_name not in name:
                    continue
                # Avoid spurious match with ".up_proj".
                if "altup_projections" in name:
                    continue
                name = name.replace(shard_name, param_name)
                # Skip loading extra bias for GPTQ models.
                if name.endswith(".bias") and name not in params_dict:
                    continue
                if is_pp_missing_parameter(name, self):
                    continue
                param = params_dict[name]
                weight_loader = param.weight_loader
                weight_loader(param, loaded_weight, shard_id)
                break
            else:
                # Skip loading extra bias for GPTQ models.
                if name.endswith(".bias") and name not in params_dict:
                    continue
                # Remapping the name of FP8 kv-scale.
                name = maybe_remap_kv_scale_name(name, params_dict)
                if name is None:
                    continue
                if is_pp_missing_parameter(name, self):
                    continue
                param = params_dict[name]
                weight_loader = getattr(param, "weight_loader",
                                        default_weight_loader)
                weight_loader(param, loaded_weight)
            loaded_params.add(name)

        return loaded_params

altup_projections instance-attribute 

altup_projections = ModuleList(
    [
        ColumnParallelLinear(
            hidden_size,
            hidden_size,
            bias=False,
            gather_output=True,
            return_bias=False,
            prefix=f"{prefix}.{idx - 1}.altup_projections",
        )
        for idx in range(1, altup_num_inputs)
    ]
)

altup_unembed_projections instance-attribute 

altup_unembed_projections = ModuleList(
    [
        ColumnParallelLinear(
            hidden_size,
            hidden_size,
            bias=False,
            gather_output=True,
            return_bias=False,
            prefix=f"{prefix}.{idx - 1}.altup_unembed_projections",
        )
        for idx in range(1, altup_num_inputs)
    ]
)

config instance-attribute 

config = config

embed_scale instance-attribute 

embed_scale = tensor(hidden_size ** 0.5, dtype=dtype)

embed_scale_per_layer instance-attribute 

embed_scale_per_layer = tensor(
    hidden_size_per_layer_input**0.5, dtype=dtype
)

embed_tokens instance-attribute 

embed_tokens = VocabParallelEmbedding(
    vocab_size, hidden_size, prefix=f"{prefix}.embed_tokens"
)

embed_tokens_per_layer instance-attribute 

embed_tokens_per_layer = VocabParallelEmbedding(
    vocab_size_per_layer_input,
    num_hidden_layers * hidden_size_per_layer_input,
    prefix=f"{prefix}.per_layer_embed_tokens",
)

eps instance-attribute 

eps = tensor(min)

norm instance-attribute 

norm = RMSNorm(hidden_size, eps=rms_norm_eps)

per_layer_input_scale instance-attribute 

per_layer_input_scale = to(dtype)

per_layer_model_projection instance-attribute 

per_layer_model_projection = ColumnParallelLinear(
    hidden_size,
    num_hidden_layers * hidden_size_per_layer_input,
    bias=False,
    gather_output=True,
    return_bias=False,
    prefix=f"{prefix}.per_layer_model_projection",
)

per_layer_projection_norm instance-attribute 

per_layer_projection_norm = RMSNorm(
    hidden_size=hidden_size_per_layer_input,
    eps=rms_norm_eps,
)

per_layer_projection_scale instance-attribute 

per_layer_projection_scale = tensor(
    hidden_size**0.5, dtype=dtype
)

quant_config instance-attribute 

quant_config = quant_config

__init__ 

__init__(*, vllm_config: VllmConfig, prefix: str = '')
Source code in vllm/model_executor/models/gemma3n.py 
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
def __init__(self, *, vllm_config: VllmConfig, prefix: str = ""):
    super().__init__()
    config = vllm_config.model_config.hf_config.text_config
    cache_config = vllm_config.cache_config
    quant_config = vllm_config.quant_config
    self.config = config
    self.quant_config = quant_config

    self.embed_tokens = VocabParallelEmbedding(
        config.vocab_size,
        config.hidden_size,
        prefix=f"{prefix}.embed_tokens",
    )
    self.embed_scale = torch.tensor(
        config.hidden_size**0.5,
        dtype=self.embed_tokens.weight.dtype,
    )
    self.embed_tokens_per_layer = VocabParallelEmbedding(
        config.vocab_size_per_layer_input,
        config.num_hidden_layers * config.hidden_size_per_layer_input,
        prefix=f"{prefix}.per_layer_embed_tokens",
    )
    self.embed_scale_per_layer = torch.tensor(
        config.hidden_size_per_layer_input**0.5,
        dtype=self.embed_tokens.weight.dtype,
    )
    self.per_layer_model_projection = ColumnParallelLinear(
        config.hidden_size,
        config.num_hidden_layers * config.hidden_size_per_layer_input,
        bias=False,
        gather_output=True,
        return_bias=False,
        prefix=f"{prefix}.per_layer_model_projection",
    )
    self.per_layer_projection_norm = RMSNorm(
        hidden_size=config.hidden_size_per_layer_input,
        eps=config.rms_norm_eps,
    )
    self.per_layer_input_scale = torch.rsqrt(torch.tensor(2.0)).to(
        self.embed_tokens.weight.dtype)
    self.per_layer_projection_scale = torch.tensor(
        config.hidden_size**0.5,
        dtype=self.embed_tokens.weight.dtype,
    )
    self.altup_projections = nn.ModuleList([
        ColumnParallelLinear(
            config.hidden_size,
            config.hidden_size,
            bias=False,
            gather_output=True,
            return_bias=False,
            prefix=f"{prefix}.{idx-1}.altup_projections",
        ) for idx in range(1, self.config.altup_num_inputs)
    ])
    self.altup_unembed_projections = nn.ModuleList([
        ColumnParallelLinear(
            config.hidden_size,
            config.hidden_size,
            bias=False,
            gather_output=True,
            return_bias=False,
            prefix=f"{prefix}.{idx-1}.altup_unembed_projections",
        ) for idx in range(1, self.config.altup_num_inputs)
    ])

    # Transformer blocks.
    self.start_layer, self.end_layer, self.layers = make_layers(
        config.num_hidden_layers,
        lambda prefix: Gemma3nDecoderLayer(
            config, cache_config, quant_config, prefix=prefix),
        prefix=f"{prefix}.layers")
    self.norm = RMSNorm(
        config.hidden_size,
        eps=config.rms_norm_eps,
    )
    self.eps = torch.tensor(torch.finfo().min)

forward 

forward(
    input_ids: Optional[Tensor],
    positions: Tensor,
    inputs_embeds: Optional[Tensor] = None,
    **kwargs,
) -> Union[Tensor, IntermediateTensors]
Source code in vllm/model_executor/models/gemma3n.py 
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
def forward(
    self,
    input_ids: Optional[torch.Tensor],
    positions: torch.Tensor,
    inputs_embeds: Optional[torch.Tensor] = None,
    **kwargs,
) -> Union[torch.Tensor, IntermediateTensors]:
    if inputs_embeds is not None:
        hidden_states_0 = inputs_embeds
    else:
        hidden_states_0 = self.get_input_embeddings(input_ids)

    # Per layer inputs.
    if input_ids is None:
        raise ValueError("Passing None for input ids is not supported.")
    per_layer_inputs = self.get_per_layer_input_embeddings(input_ids)
    per_layer_inputs = per_layer_inputs.reshape(
        -1, self.config.num_hidden_layers,
        self.config.hidden_size_per_layer_input)
    per_layer_projection = self.per_layer_model_projection(hidden_states_0)
    per_layer_projection = per_layer_projection.reshape(
        *hidden_states_0.shape[:-1],
        self.config.num_hidden_layers,
        self.config.hidden_size_per_layer_input,
    )
    per_layer_projection = self.per_layer_projection_norm(
        per_layer_projection)
    per_layer_inputs = per_layer_projection + per_layer_inputs
    per_layer_inputs *= self.per_layer_input_scale

    # Altup embed.
    hidden_states = [hidden_states_0] * self.config.altup_num_inputs
    target_magnitude = torch.mean(hidden_states_0**2, dim=-1,
                                  keepdim=True)**0.5
    for i in range(1, self.config.altup_num_inputs):
        hidden_states[i] = self.altup_projections[i - 1](hidden_states[i])
        new_magnitude = torch.mean(hidden_states[i]**2,
                                   dim=-1,
                                   keepdim=True)**0.5
        hidden_states[i] *= target_magnitude / torch.maximum(
            new_magnitude, self.eps)
    hidden_states = torch.stack(hidden_states, dim=0)

    # Transformer blocks.
    for layer_idx, layer in enumerate(self.layers):
        # [altup_num_inputs, num_tokens, hidden_size]
        hidden_states = layer(
            positions=positions,
            hidden_states=hidden_states,
            per_layer_input=per_layer_inputs[:, layer_idx, :],
            **kwargs,
        )

    # Altup unembed.
    target_magnitude = torch.mean(hidden_states[0]**2,
                                  dim=-1,
                                  keepdim=True)**0.5
    for i in range(1, self.config.altup_num_inputs):
        hidden_states[i] = self.altup_unembed_projections[i - 1](
            hidden_states[i])
        new_magnitude = torch.mean(hidden_states[i]**2,
                                   dim=-1,
                                   keepdim=True)**0.5
        hidden_states[i] *= target_magnitude / torch.maximum(
            new_magnitude, self.eps)
    # [altup_num_inputs,num_tokens,hidden_size] -> [num_tokens,hidden_size]
    hidden_states = torch.mean(hidden_states, dim=0)

    return self.norm(hidden_states)

get_input_embeddings 

get_input_embeddings(input_ids: Tensor) -> Tensor
Source code in vllm/model_executor/models/gemma3n.py 
590
591
def get_input_embeddings(self, input_ids: torch.Tensor) -> torch.Tensor:
    return self.embed_tokens(input_ids) * self.embed_scale

get_per_layer_input_embeddings 

get_per_layer_input_embeddings(input_ids: Tensor) -> Tensor
Source code in vllm/model_executor/models/gemma3n.py 
593
594
595
596
597
598
599
600
601
602
603
def get_per_layer_input_embeddings(
        self, input_ids: torch.Tensor) -> torch.Tensor:
    # Deal with the fact that vocab_size_per_layer_input < vocab_size
    # which causes us to have some out of vocab tokens by setting
    # those token ids to 0. This matches the HF implementation.
    per_layer_inputs_mask = torch.logical_and(
        input_ids >= 0, input_ids < self.config.vocab_size_per_layer_input)
    per_layer_inputs_tokens = torch.where(per_layer_inputs_mask, input_ids,
                                          torch.zeros_like(input_ids))
    return self.embed_tokens_per_layer(
        per_layer_inputs_tokens) * self.embed_scale_per_layer

load_weights 

load_weights(
    weights: Iterable[tuple[str, Tensor]],
) -> set[str]
Source code in vllm/model_executor/models/gemma3n.py 
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
def load_weights(self, weights: Iterable[tuple[str,
                                               torch.Tensor]]) -> set[str]:
    stacked_params_mapping = [
        # (param_name, shard_name, shard_id)
        ("qkv_proj", "q_proj", "q"),
        ("qkv_proj", "k_proj", "k"),
        ("qkv_proj", "v_proj", "v"),
        ("gate_up_proj", "gate_proj", 0),
        ("gate_up_proj", "up_proj", 1),
    ]
    params_dict = dict(self.named_parameters())
    loaded_params: set[str] = set()
    for name, loaded_weight in weights:
        if (self.quant_config is not None and
            (scale_name := self.quant_config.get_cache_scale(name))):
            # Loading kv cache scales for compressed-tensors quantization
            param = params_dict[scale_name]
            weight_loader = getattr(param, "weight_loader",
                                    default_weight_loader)
            loaded_weight = loaded_weight[0]
            weight_loader(param, loaded_weight)
            loaded_params.add(scale_name)
            continue
        for (param_name, shard_name, shard_id) in stacked_params_mapping:
            if shard_name not in name:
                continue
            # Avoid spurious match with ".up_proj".
            if "altup_projections" in name:
                continue
            name = name.replace(shard_name, param_name)
            # Skip loading extra bias for GPTQ models.
            if name.endswith(".bias") and name not in params_dict:
                continue
            if is_pp_missing_parameter(name, self):
                continue
            param = params_dict[name]
            weight_loader = param.weight_loader
            weight_loader(param, loaded_weight, shard_id)
            break
        else:
            # Skip loading extra bias for GPTQ models.
            if name.endswith(".bias") and name not in params_dict:
                continue
            # Remapping the name of FP8 kv-scale.
            name = maybe_remap_kv_scale_name(name, params_dict)
            if name is None:
                continue
            if is_pp_missing_parameter(name, self):
                continue
            param = params_dict[name]
            weight_loader = getattr(param, "weight_loader",
                                    default_weight_loader)
            weight_loader(param, loaded_weight)
        loaded_params.add(name)

    return loaded_params