vllm.model_executor.models.falcon

PyTorch Falcon model.

FalconConfig module-attribute

FalconConfig = Union[FalconConfig, RWConfig]

FalconAttention

Bases: Module

Source code in vllm/model_executor/models/falcon.py

class FalconAttention(nn.Module):

    def __init__(
        self,
        config: FalconConfig,
        cache_config: Optional[CacheConfig] = None,
        quant_config: Optional[QuantizationConfig] = None,
        prefix: str = "",
    ):
        super().__init__()

        self.hidden_size = config.hidden_size
        tp_size = get_tensor_model_parallel_world_size()

        self.total_num_heads = config.num_attention_heads
        assert self.total_num_heads % tp_size == 0
        self.num_heads = self.total_num_heads // tp_size
        self.head_dim = self.hidden_size // self.total_num_heads
        assert self.head_dim * self.total_num_heads == self.hidden_size

        self.new_decoder_architecture = config.new_decoder_architecture
        self.multi_query = config.multi_query

        if self.new_decoder_architecture:
            self.total_num_kv_heads = config.num_kv_heads
        elif self.multi_query:
            self.total_num_kv_heads = 1
        else:
            self.total_num_kv_heads = self.total_num_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.query_key_value = QKVParallelLinear(
            self.hidden_size,
            self.head_dim,
            self.total_num_heads,
            self.total_num_kv_heads,
            bias=config.bias,
            skip_bias_add=True,
            quant_config=quant_config,
        )
        self.q_size = self.num_heads * self.head_dim
        self.kv_size = self.num_kv_heads * self.head_dim

        # Layer-wise attention scaling
        self.inv_norm_factor = 1.0 / math.sqrt(self.head_dim)
        self.reduce_row_parallel_results = not (config.new_decoder_architecture
                                                or config.parallel_attn)
        self.dense = RowParallelLinear(
            self.hidden_size,
            self.hidden_size,
            bias=config.bias,
            skip_bias_add=True,
            quant_config=quant_config,
            reduce_results=self.reduce_row_parallel_results)

        self.use_rotary = config.rotary
        self.use_alibi = config.alibi
        assert not (self.use_rotary and self.use_alibi), (
            "Rotary and alibi are mutually exclusive.")

        if self.use_rotary:
            rope_theta = getattr(config, "rope_theta", 10000)
            max_position_embeddings = getattr(config,
                                              "max_position_embeddings", 8192)
            self.rotary_emb = get_rope(
                self.head_dim,
                rotary_dim=self.head_dim,
                max_position=max_position_embeddings,
                base=rope_theta,
            )
            self.attn = Attention(self.num_heads,
                                  self.head_dim,
                                  self.inv_norm_factor,
                                  num_kv_heads=self.num_kv_heads,
                                  quant_config=quant_config,
                                  prefix=f"{prefix}.attn")
        elif self.use_alibi:
            tp_rank = get_tensor_model_parallel_rank()
            head_start = tp_rank * self.num_heads
            head_end = (tp_rank + 1) * self.num_heads
            alibi_slopes = (_get_alibi_slopes(self.total_num_heads) *
                            self.inv_norm_factor)
            alibi_slopes = alibi_slopes[head_start:head_end].tolist()
            self.attn = Attention(self.num_heads,
                                  self.head_dim,
                                  self.inv_norm_factor,
                                  num_kv_heads=self.num_kv_heads,
                                  alibi_slopes=alibi_slopes,
                                  quant_config=quant_config,
                                  prefix=f"{prefix}.attn")
        else:
            self.attn = Attention(self.num_heads,
                                  self.head_dim,
                                  scale=self.inv_norm_factor,
                                  num_kv_heads=self.num_kv_heads,
                                  cache_config=cache_config,
                                  quant_config=quant_config,
                                  prefix=f"{prefix}.attn")

    def forward(
        self,
        positions: torch.Tensor,
        hidden_states: torch.Tensor,
    ) -> torch.Tensor:
        qkv, bias = self.query_key_value(hidden_states)
        if bias is not None:
            qkv += bias
        q, k, v = qkv.split([self.q_size, self.kv_size, self.kv_size], dim=-1)
        if self.use_rotary:
            q, k = self.rotary_emb(positions, q, k)
        attn_output = self.attn(q, k, v)
        attn_output, bias = self.dense(attn_output)
        return attn_output, bias

attn instance-attribute

attn = Attention(
    num_heads,
    head_dim,
    inv_norm_factor,
    num_kv_heads=num_kv_heads,
    quant_config=quant_config,
    prefix=f"{prefix}.attn",
)

dense instance-attribute

dense = RowParallelLinear(
    hidden_size,
    hidden_size,
    bias=bias,
    skip_bias_add=True,
    quant_config=quant_config,
    reduce_results=reduce_row_parallel_results,
)

head_dim instance-attribute

head_dim = hidden_size // total_num_heads

hidden_size instance-attribute

hidden_size = hidden_size

inv_norm_factor instance-attribute

inv_norm_factor = 1.0 / sqrt(head_dim)

kv_size instance-attribute

kv_size = num_kv_heads * head_dim

multi_query instance-attribute

multi_query = multi_query

new_decoder_architecture instance-attribute

new_decoder_architecture = new_decoder_architecture

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)

q_size instance-attribute

q_size = num_heads * head_dim

query_key_value instance-attribute

query_key_value = QKVParallelLinear(
    hidden_size,
    head_dim,
    total_num_heads,
    total_num_kv_heads,
    bias=bias,
    skip_bias_add=True,
    quant_config=quant_config,
)

reduce_row_parallel_results instance-attribute

reduce_row_parallel_results = (
    not new_decoder_architecture or parallel_attn
)

rotary_emb instance-attribute

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

total_num_heads instance-attribute

total_num_heads = num_attention_heads

total_num_kv_heads instance-attribute

total_num_kv_heads = num_kv_heads

use_alibi instance-attribute

use_alibi = alibi

use_rotary instance-attribute

use_rotary = rotary

__init__

__init__(
    config: FalconConfig,
    cache_config: Optional[CacheConfig] = None,
    quant_config: Optional[QuantizationConfig] = None,
    prefix: str = "",
)

Source code in vllm/model_executor/models/falcon.py

def __init__(
    self,
    config: FalconConfig,
    cache_config: Optional[CacheConfig] = None,
    quant_config: Optional[QuantizationConfig] = None,
    prefix: str = "",
):
    super().__init__()

    self.hidden_size = config.hidden_size
    tp_size = get_tensor_model_parallel_world_size()

    self.total_num_heads = config.num_attention_heads
    assert self.total_num_heads % tp_size == 0
    self.num_heads = self.total_num_heads // tp_size
    self.head_dim = self.hidden_size // self.total_num_heads
    assert self.head_dim * self.total_num_heads == self.hidden_size

    self.new_decoder_architecture = config.new_decoder_architecture
    self.multi_query = config.multi_query

    if self.new_decoder_architecture:
        self.total_num_kv_heads = config.num_kv_heads
    elif self.multi_query:
        self.total_num_kv_heads = 1
    else:
        self.total_num_kv_heads = self.total_num_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.query_key_value = QKVParallelLinear(
        self.hidden_size,
        self.head_dim,
        self.total_num_heads,
        self.total_num_kv_heads,
        bias=config.bias,
        skip_bias_add=True,
        quant_config=quant_config,
    )
    self.q_size = self.num_heads * self.head_dim
    self.kv_size = self.num_kv_heads * self.head_dim

    # Layer-wise attention scaling
    self.inv_norm_factor = 1.0 / math.sqrt(self.head_dim)
    self.reduce_row_parallel_results = not (config.new_decoder_architecture
                                            or config.parallel_attn)
    self.dense = RowParallelLinear(
        self.hidden_size,
        self.hidden_size,
        bias=config.bias,
        skip_bias_add=True,
        quant_config=quant_config,
        reduce_results=self.reduce_row_parallel_results)

    self.use_rotary = config.rotary
    self.use_alibi = config.alibi
    assert not (self.use_rotary and self.use_alibi), (
        "Rotary and alibi are mutually exclusive.")

    if self.use_rotary:
        rope_theta = getattr(config, "rope_theta", 10000)
        max_position_embeddings = getattr(config,
                                          "max_position_embeddings", 8192)
        self.rotary_emb = get_rope(
            self.head_dim,
            rotary_dim=self.head_dim,
            max_position=max_position_embeddings,
            base=rope_theta,
        )
        self.attn = Attention(self.num_heads,
                              self.head_dim,
                              self.inv_norm_factor,
                              num_kv_heads=self.num_kv_heads,
                              quant_config=quant_config,
                              prefix=f"{prefix}.attn")
    elif self.use_alibi:
        tp_rank = get_tensor_model_parallel_rank()
        head_start = tp_rank * self.num_heads
        head_end = (tp_rank + 1) * self.num_heads
        alibi_slopes = (_get_alibi_slopes(self.total_num_heads) *
                        self.inv_norm_factor)
        alibi_slopes = alibi_slopes[head_start:head_end].tolist()
        self.attn = Attention(self.num_heads,
                              self.head_dim,
                              self.inv_norm_factor,
                              num_kv_heads=self.num_kv_heads,
                              alibi_slopes=alibi_slopes,
                              quant_config=quant_config,
                              prefix=f"{prefix}.attn")
    else:
        self.attn = Attention(self.num_heads,
                              self.head_dim,
                              scale=self.inv_norm_factor,
                              num_kv_heads=self.num_kv_heads,
                              cache_config=cache_config,
                              quant_config=quant_config,
                              prefix=f"{prefix}.attn")

forward

forward(positions: Tensor, hidden_states: Tensor) -> Tensor

Source code in vllm/model_executor/models/falcon.py

def forward(
    self,
    positions: torch.Tensor,
    hidden_states: torch.Tensor,
) -> torch.Tensor:
    qkv, bias = self.query_key_value(hidden_states)
    if bias is not None:
        qkv += bias
    q, k, v = qkv.split([self.q_size, self.kv_size, self.kv_size], dim=-1)
    if self.use_rotary:
        q, k = self.rotary_emb(positions, q, k)
    attn_output = self.attn(q, k, v)
    attn_output, bias = self.dense(attn_output)
    return attn_output, bias

FalconDecoderLayer

Bases: Module

Source code in vllm/model_executor/models/falcon.py

class FalconDecoderLayer(nn.Module):

    def __init__(
        self,
        config: FalconConfig,
        cache_config: Optional[CacheConfig] = None,
        quant_config: Optional[QuantizationConfig] = None,
        prefix: str = "",
    ):
        super().__init__()
        hidden_size = config.hidden_size
        self.num_heads = config.num_attention_heads
        self.self_attention = FalconAttention(
            config,
            cache_config,
            quant_config,
            prefix=f"{prefix}.self_attention")
        self.mlp = FalconMLP(config, quant_config)
        self.config = config

        if (not hasattr(config, "num_ln_in_parallel_attn")):
            config.num_ln_in_parallel_attn = None

        if (config.num_ln_in_parallel_attn is None
                and config.new_decoder_architecture):
            config.num_ln_in_parallel_attn = 2

        if not config.parallel_attn:
            self.post_attention_layernorm = LayerNorm(
                hidden_size, eps=config.layer_norm_epsilon)
            self.input_layernorm = LayerNorm(hidden_size,
                                             eps=config.layer_norm_epsilon)
        else:
            if config.num_ln_in_parallel_attn == 2:
                # The layer norm before self-attention
                self.ln_attn = LayerNorm(hidden_size,
                                         eps=config.layer_norm_epsilon)
                # The layer norm before the MLP
                self.ln_mlp = LayerNorm(hidden_size,
                                        eps=config.layer_norm_epsilon)
            else:
                self.input_layernorm = LayerNorm(hidden_size,
                                                 eps=config.layer_norm_epsilon)

        self.reduce_row_parallel_results = not (config.new_decoder_architecture
                                                or config.parallel_attn)

    def forward(
        self,
        positions: torch.Tensor,
        hidden_states: torch.Tensor,
    ) -> torch.Tensor:
        residual = hidden_states

        if self.config.num_ln_in_parallel_attn == 2:
            attention_layernorm_out = self.ln_attn(hidden_states)
            mlp_layernorm_out = self.ln_mlp(hidden_states)
        else:
            attention_layernorm_out = self.input_layernorm(hidden_states)

        # Self attention.
        attention_output, attention_bias = self.self_attention(
            positions=positions,
            hidden_states=attention_layernorm_out,
        )
        if self.reduce_row_parallel_results and attention_bias is not None:
            attention_output += attention_bias

        if not self.config.new_decoder_architecture:
            if self.config.parallel_attn:
                mlp_layernorm_out = attention_layernorm_out
            else:
                residual += attention_output
                mlp_layernorm_out = self.post_attention_layernorm(residual)

        if (self.config.new_decoder_architecture and self.config.parallel_attn
                and self.config.num_ln_in_parallel_attn == 1):
            mlp_layernorm_out = attention_layernorm_out

        # MLP.
        mlp_output, mlp_bias = self.mlp(mlp_layernorm_out)
        if self.reduce_row_parallel_results and mlp_bias is not None:
            mlp_output += mlp_bias

        if not self.reduce_row_parallel_results:
            # When MLP and Attention layers are parallel, we can use
            # only one all-reduce operator to reduce the results from
            # both MLP and Attention layers.
            mlp_output += attention_output
            mlp_output = tensor_model_parallel_all_reduce(mlp_output)
            if attention_bias is not None:
                mlp_output += attention_bias
            if mlp_bias is not None:
                mlp_output += mlp_bias

        output = mlp_output + residual
        return output

config instance-attribute

config = config

input_layernorm instance-attribute

input_layernorm = LayerNorm(
    hidden_size, eps=layer_norm_epsilon
)

ln_attn instance-attribute

ln_attn = LayerNorm(hidden_size, eps=layer_norm_epsilon)

ln_mlp instance-attribute

ln_mlp = LayerNorm(hidden_size, eps=layer_norm_epsilon)

mlp instance-attribute

mlp = FalconMLP(config, quant_config)

num_heads instance-attribute

num_heads = num_attention_heads

post_attention_layernorm instance-attribute

post_attention_layernorm = LayerNorm(
    hidden_size, eps=layer_norm_epsilon
)

reduce_row_parallel_results instance-attribute

reduce_row_parallel_results = (
    not new_decoder_architecture or parallel_attn
)

self_attention instance-attribute

self_attention = FalconAttention(
    config,
    cache_config,
    quant_config,
    prefix=f"{prefix}.self_attention",
)

__init__

__init__(
    config: FalconConfig,
    cache_config: Optional[CacheConfig] = None,
    quant_config: Optional[QuantizationConfig] = None,
    prefix: str = "",
)

Source code in vllm/model_executor/models/falcon.py

def __init__(
    self,
    config: FalconConfig,
    cache_config: Optional[CacheConfig] = None,
    quant_config: Optional[QuantizationConfig] = None,
    prefix: str = "",
):
    super().__init__()
    hidden_size = config.hidden_size
    self.num_heads = config.num_attention_heads
    self.self_attention = FalconAttention(
        config,
        cache_config,
        quant_config,
        prefix=f"{prefix}.self_attention")
    self.mlp = FalconMLP(config, quant_config)
    self.config = config

    if (not hasattr(config, "num_ln_in_parallel_attn")):
        config.num_ln_in_parallel_attn = None

    if (config.num_ln_in_parallel_attn is None
            and config.new_decoder_architecture):
        config.num_ln_in_parallel_attn = 2

    if not config.parallel_attn:
        self.post_attention_layernorm = LayerNorm(
            hidden_size, eps=config.layer_norm_epsilon)
        self.input_layernorm = LayerNorm(hidden_size,
                                         eps=config.layer_norm_epsilon)
    else:
        if config.num_ln_in_parallel_attn == 2:
            # The layer norm before self-attention
            self.ln_attn = LayerNorm(hidden_size,
                                     eps=config.layer_norm_epsilon)
            # The layer norm before the MLP
            self.ln_mlp = LayerNorm(hidden_size,
                                    eps=config.layer_norm_epsilon)
        else:
            self.input_layernorm = LayerNorm(hidden_size,
                                             eps=config.layer_norm_epsilon)

    self.reduce_row_parallel_results = not (config.new_decoder_architecture
                                            or config.parallel_attn)

forward

forward(positions: Tensor, hidden_states: Tensor) -> Tensor

Source code in vllm/model_executor/models/falcon.py

def forward(
    self,
    positions: torch.Tensor,
    hidden_states: torch.Tensor,
) -> torch.Tensor:
    residual = hidden_states

    if self.config.num_ln_in_parallel_attn == 2:
        attention_layernorm_out = self.ln_attn(hidden_states)
        mlp_layernorm_out = self.ln_mlp(hidden_states)
    else:
        attention_layernorm_out = self.input_layernorm(hidden_states)

    # Self attention.
    attention_output, attention_bias = self.self_attention(
        positions=positions,
        hidden_states=attention_layernorm_out,
    )
    if self.reduce_row_parallel_results and attention_bias is not None:
        attention_output += attention_bias

    if not self.config.new_decoder_architecture:
        if self.config.parallel_attn:
            mlp_layernorm_out = attention_layernorm_out
        else:
            residual += attention_output
            mlp_layernorm_out = self.post_attention_layernorm(residual)

    if (self.config.new_decoder_architecture and self.config.parallel_attn
            and self.config.num_ln_in_parallel_attn == 1):
        mlp_layernorm_out = attention_layernorm_out

    # MLP.
    mlp_output, mlp_bias = self.mlp(mlp_layernorm_out)
    if self.reduce_row_parallel_results and mlp_bias is not None:
        mlp_output += mlp_bias

    if not self.reduce_row_parallel_results:
        # When MLP and Attention layers are parallel, we can use
        # only one all-reduce operator to reduce the results from
        # both MLP and Attention layers.
        mlp_output += attention_output
        mlp_output = tensor_model_parallel_all_reduce(mlp_output)
        if attention_bias is not None:
            mlp_output += attention_bias
        if mlp_bias is not None:
            mlp_output += mlp_bias

    output = mlp_output + residual
    return output

FalconForCausalLM

Bases: Module, SupportsPP

Source code in vllm/model_executor/models/falcon.py

class FalconForCausalLM(nn.Module, SupportsPP):
    packed_modules_mapping = {
        "query_key_value": ["query_key_value"],
    }

    def __init__(self, *, vllm_config: VllmConfig, prefix: str = ""):
        super().__init__()
        config = vllm_config.model_config.hf_config
        quant_config = vllm_config.quant_config
        self.config = config
        self.quant_config = quant_config
        self.transformer = FalconModel(vllm_config=vllm_config,
                                       prefix=maybe_prefix(
                                           prefix, "transformer"))
        # only Falcon-11B doesn't share lm_head weight with word embeddings
        # and previous Falcon model doesn't have tie_word_embeddings config
        # so we set tie_word_embeddings to True by default
        self.tie_word_embeddings = (config.tie_word_embeddings
                                    if config.tie_word_embeddings is not None
                                    else True)
        if self.tie_word_embeddings:
            self.lm_head = self.transformer.word_embeddings
        else:
            self.lm_head = ParallelLMHead(
                config.vocab_size,
                config.hidden_size,
                quant_config=quant_config,
            )
        self.logits_processor = LogitsProcessor(config.vocab_size)
        self.make_empty_intermediate_tensors = (
            self.transformer.make_empty_intermediate_tensors)

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

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

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

    def load_weights(self, weights: Iterable[tuple[str,
                                                   torch.Tensor]]) -> set[str]:
        loader = AutoWeightsLoader(
            self,
            skip_prefixes=(["lm_head."]
                           if self.config.tie_word_embeddings else None),
        )
        return loader.load_weights(weights)

config instance-attribute

config = config

lm_head instance-attribute

lm_head = word_embeddings

logits_processor instance-attribute

logits_processor = LogitsProcessor(vocab_size)

make_empty_intermediate_tensors instance-attribute

make_empty_intermediate_tensors = (
    make_empty_intermediate_tensors
)

packed_modules_mapping class-attribute instance-attribute

packed_modules_mapping = {
    "query_key_value": ["query_key_value"]
}

quant_config instance-attribute

quant_config = quant_config

tie_word_embeddings instance-attribute

tie_word_embeddings = (
    tie_word_embeddings
    if tie_word_embeddings is not None
    else True
)

transformer instance-attribute

transformer = FalconModel(
    vllm_config=vllm_config,
    prefix=maybe_prefix(prefix, "transformer"),
)

__init__

__init__(*, vllm_config: VllmConfig, prefix: str = '')

Source code in vllm/model_executor/models/falcon.py

def __init__(self, *, vllm_config: VllmConfig, prefix: str = ""):
    super().__init__()
    config = vllm_config.model_config.hf_config
    quant_config = vllm_config.quant_config
    self.config = config
    self.quant_config = quant_config
    self.transformer = FalconModel(vllm_config=vllm_config,
                                   prefix=maybe_prefix(
                                       prefix, "transformer"))
    # only Falcon-11B doesn't share lm_head weight with word embeddings
    # and previous Falcon model doesn't have tie_word_embeddings config
    # so we set tie_word_embeddings to True by default
    self.tie_word_embeddings = (config.tie_word_embeddings
                                if config.tie_word_embeddings is not None
                                else True)
    if self.tie_word_embeddings:
        self.lm_head = self.transformer.word_embeddings
    else:
        self.lm_head = ParallelLMHead(
            config.vocab_size,
            config.hidden_size,
            quant_config=quant_config,
        )
    self.logits_processor = LogitsProcessor(config.vocab_size)
    self.make_empty_intermediate_tensors = (
        self.transformer.make_empty_intermediate_tensors)

compute_logits

compute_logits(
    hidden_states: Tensor,
    sampling_metadata: SamplingMetadata,
) -> Optional[Tensor]

Source code in vllm/model_executor/models/falcon.py

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

forward

forward(
    input_ids: LongTensor,
    positions: Tensor,
    intermediate_tensors: Optional[
        IntermediateTensors
    ] = None,
    inputs_embeds: Optional[Tensor] = None,
) -> Tensor

Source code in vllm/model_executor/models/falcon.py

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

get_input_embeddings

get_input_embeddings(input_ids: Tensor) -> Tensor

Source code in vllm/model_executor/models/falcon.py

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

load_weights

load_weights(
    weights: Iterable[tuple[str, Tensor]],
) -> set[str]

Source code in vllm/model_executor/models/falcon.py

def load_weights(self, weights: Iterable[tuple[str,
                                               torch.Tensor]]) -> set[str]:
    loader = AutoWeightsLoader(
        self,
        skip_prefixes=(["lm_head."]
                       if self.config.tie_word_embeddings else None),
    )
    return loader.load_weights(weights)

FalconMLP

Bases: Module

Source code in vllm/model_executor/models/falcon.py

class FalconMLP(nn.Module):

    def __init__(
        self,
        config: FalconConfig,
        quant_config: Optional[QuantizationConfig] = None,
    ):
        super().__init__()
        hidden_size = config.hidden_size

        self.dense_h_to_4h = ColumnParallelLinear(hidden_size,
                                                  4 * hidden_size,
                                                  bias=config.bias,
                                                  skip_bias_add=True,
                                                  quant_config=quant_config)
        self.act = get_act_fn("gelu")
        self.reduce_row_parallel_results = not (config.new_decoder_architecture
                                                or config.parallel_attn)
        self.dense_4h_to_h = RowParallelLinear(
            4 * hidden_size,
            hidden_size,
            bias=config.bias,
            skip_bias_add=True,
            reduce_results=self.reduce_row_parallel_results,
            quant_config=quant_config)

    def forward(self, x: torch.Tensor) -> torch.Tensor:
        # NOTE(zhuohan): Following huggingface, we do not fuse bias add here.
        x, bias = self.dense_h_to_4h(x)
        if bias is not None:
            x += bias
        x = self.act(x)
        x, bias = self.dense_4h_to_h(x)
        return x, bias

act instance-attribute

act = get_act_fn('gelu')

dense_4h_to_h instance-attribute

dense_4h_to_h = RowParallelLinear(
    4 * hidden_size,
    hidden_size,
    bias=bias,
    skip_bias_add=True,
    reduce_results=reduce_row_parallel_results,
    quant_config=quant_config,
)

dense_h_to_4h instance-attribute

dense_h_to_4h = ColumnParallelLinear(
    hidden_size,
    4 * hidden_size,
    bias=bias,
    skip_bias_add=True,
    quant_config=quant_config,
)

reduce_row_parallel_results instance-attribute

reduce_row_parallel_results = (
    not new_decoder_architecture or parallel_attn
)

__init__

__init__(
    config: FalconConfig,
    quant_config: Optional[QuantizationConfig] = None,
)

Source code in vllm/model_executor/models/falcon.py

def __init__(
    self,
    config: FalconConfig,
    quant_config: Optional[QuantizationConfig] = None,
):
    super().__init__()
    hidden_size = config.hidden_size

    self.dense_h_to_4h = ColumnParallelLinear(hidden_size,
                                              4 * hidden_size,
                                              bias=config.bias,
                                              skip_bias_add=True,
                                              quant_config=quant_config)
    self.act = get_act_fn("gelu")
    self.reduce_row_parallel_results = not (config.new_decoder_architecture
                                            or config.parallel_attn)
    self.dense_4h_to_h = RowParallelLinear(
        4 * hidden_size,
        hidden_size,
        bias=config.bias,
        skip_bias_add=True,
        reduce_results=self.reduce_row_parallel_results,
        quant_config=quant_config)

forward

forward(x: Tensor) -> Tensor

Source code in vllm/model_executor/models/falcon.py

def forward(self, x: torch.Tensor) -> torch.Tensor:
    # NOTE(zhuohan): Following huggingface, we do not fuse bias add here.
    x, bias = self.dense_h_to_4h(x)
    if bias is not None:
        x += bias
    x = self.act(x)
    x, bias = self.dense_4h_to_h(x)
    return x, bias

FalconModel

Bases: Module

Source code in vllm/model_executor/models/falcon.py

@support_torch_compile
class FalconModel(nn.Module):

    def __init__(self, *, vllm_config: VllmConfig, prefix: str = ""):
        super().__init__()

        config = vllm_config.model_config.hf_config
        cache_config = vllm_config.cache_config
        quant_config = vllm_config.quant_config

        self.config = config
        self.embed_dim = config.hidden_size
        self.num_heads = config.num_attention_heads
        self.use_alibi = config.alibi

        # Embedding + LN Embedding
        self.word_embeddings = VocabParallelEmbedding(
            config.vocab_size,
            self.embed_dim,
        )

        # Transformer blocks
        self.start_layer, self.end_layer, self.h = make_layers(
            config.num_hidden_layers,
            lambda prefix: FalconDecoderLayer(
                config, cache_config, quant_config, prefix=prefix),
            prefix=f"{prefix}.h")

        # Final Layer Norm
        self.ln_f = LayerNorm(self.embed_dim, eps=config.layer_norm_epsilon)
        self.make_empty_intermediate_tensors = (
            make_empty_intermediate_tensors_factory(["hidden_states"],
                                                    config.hidden_size))

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

    def forward(
        self,
        input_ids: torch.Tensor,
        positions: torch.Tensor,
        intermediate_tensors: Optional[IntermediateTensors],
        inputs_embeds: Optional[torch.Tensor] = None,
    ) -> Union[torch.Tensor, IntermediateTensors]:
        if get_pp_group().is_first_rank:
            if inputs_embeds is not None:
                hidden_states = inputs_embeds
            else:
                hidden_states = self.get_input_embeddings(input_ids)
        else:
            hidden_states = intermediate_tensors["hidden_states"]
        for layer in self.h[self.start_layer:self.end_layer]:
            hidden_states = layer(positions, hidden_states)
        if not get_pp_group().is_last_rank:
            return IntermediateTensors({"hidden_states": hidden_states})
        hidden_states = self.ln_f(hidden_states)
        return hidden_states

    def load_weights(self, weights: Iterable[tuple[str,
                                                   torch.Tensor]]) -> set[str]:
        total_num_heads = self.config.num_attention_heads
        if self.config.new_decoder_architecture:
            total_num_kv_heads = self.config.num_kv_heads
        elif self.config.multi_query:
            total_num_kv_heads = 1
        else:
            total_num_kv_heads = total_num_heads
        num_query_heads_per_kv_head = total_num_heads // total_num_kv_heads
        params_dict = dict(self.named_parameters(remove_duplicate=False))
        loaded_params: set[str] = set()
        for name, loaded_weight in weights:
            # 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]
            if "query_key_value" in name:
                output_dim = getattr(param, "output_dim", None)
                loaded_weight_shape = loaded_weight.shape
                if output_dim is not None:
                    loaded_weight = loaded_weight.view(
                        loaded_weight_shape[:output_dim] +
                        (total_num_kv_heads, num_query_heads_per_kv_head + 2,
                         -1) + loaded_weight_shape[output_dim + 1:])
                    wq = loaded_weight.narrow(
                        output_dim + 1, 0,
                        num_query_heads_per_kv_head).reshape(
                            *loaded_weight_shape[:output_dim], -1,
                            *loaded_weight_shape[output_dim + 1:])
                    wk = loaded_weight.narrow(
                        output_dim + 1, num_query_heads_per_kv_head,
                        1).reshape(*loaded_weight_shape[:output_dim], -1,
                                   *loaded_weight_shape[output_dim + 1:])
                    wv = loaded_weight.narrow(
                        output_dim + 1, num_query_heads_per_kv_head + 1,
                        1).reshape(*loaded_weight_shape[:output_dim], -1,
                                   *loaded_weight_shape[output_dim + 1:])
                    loaded_weight = torch.cat([wq, wk, wv], dim=output_dim)

            weight_loader = getattr(param, "weight_loader",
                                    default_weight_loader)
            weight_loader(param, loaded_weight)
            loaded_params.add(name)
        return loaded_params

config instance-attribute

config = config

embed_dim instance-attribute

embed_dim = hidden_size

ln_f instance-attribute

ln_f = LayerNorm(embed_dim, eps=layer_norm_epsilon)

make_empty_intermediate_tensors instance-attribute

make_empty_intermediate_tensors = (
    make_empty_intermediate_tensors_factory(
        ["hidden_states"], hidden_size
    )
)

num_heads instance-attribute

num_heads = num_attention_heads

use_alibi instance-attribute

use_alibi = alibi

word_embeddings instance-attribute

word_embeddings = VocabParallelEmbedding(
    vocab_size, embed_dim
)

__init__

__init__(*, vllm_config: VllmConfig, prefix: str = '')

Source code in vllm/model_executor/models/falcon.py

def __init__(self, *, vllm_config: VllmConfig, prefix: str = ""):
    super().__init__()

    config = vllm_config.model_config.hf_config
    cache_config = vllm_config.cache_config
    quant_config = vllm_config.quant_config

    self.config = config
    self.embed_dim = config.hidden_size
    self.num_heads = config.num_attention_heads
    self.use_alibi = config.alibi

    # Embedding + LN Embedding
    self.word_embeddings = VocabParallelEmbedding(
        config.vocab_size,
        self.embed_dim,
    )

    # Transformer blocks
    self.start_layer, self.end_layer, self.h = make_layers(
        config.num_hidden_layers,
        lambda prefix: FalconDecoderLayer(
            config, cache_config, quant_config, prefix=prefix),
        prefix=f"{prefix}.h")

    # Final Layer Norm
    self.ln_f = LayerNorm(self.embed_dim, eps=config.layer_norm_epsilon)
    self.make_empty_intermediate_tensors = (
        make_empty_intermediate_tensors_factory(["hidden_states"],
                                                config.hidden_size))

forward

forward(
    input_ids: Tensor,
    positions: Tensor,
    intermediate_tensors: Optional[IntermediateTensors],
    inputs_embeds: Optional[Tensor] = None,
) -> Union[Tensor, IntermediateTensors]

Source code in vllm/model_executor/models/falcon.py

def forward(
    self,
    input_ids: torch.Tensor,
    positions: torch.Tensor,
    intermediate_tensors: Optional[IntermediateTensors],
    inputs_embeds: Optional[torch.Tensor] = None,
) -> Union[torch.Tensor, IntermediateTensors]:
    if get_pp_group().is_first_rank:
        if inputs_embeds is not None:
            hidden_states = inputs_embeds
        else:
            hidden_states = self.get_input_embeddings(input_ids)
    else:
        hidden_states = intermediate_tensors["hidden_states"]
    for layer in self.h[self.start_layer:self.end_layer]:
        hidden_states = layer(positions, hidden_states)
    if not get_pp_group().is_last_rank:
        return IntermediateTensors({"hidden_states": hidden_states})
    hidden_states = self.ln_f(hidden_states)
    return hidden_states

get_input_embeddings

get_input_embeddings(input_ids: Tensor) -> Tensor

Source code in vllm/model_executor/models/falcon.py

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

load_weights

load_weights(
    weights: Iterable[tuple[str, Tensor]],
) -> set[str]

Source code in vllm/model_executor/models/falcon.py

def load_weights(self, weights: Iterable[tuple[str,
                                               torch.Tensor]]) -> set[str]:
    total_num_heads = self.config.num_attention_heads
    if self.config.new_decoder_architecture:
        total_num_kv_heads = self.config.num_kv_heads
    elif self.config.multi_query:
        total_num_kv_heads = 1
    else:
        total_num_kv_heads = total_num_heads
    num_query_heads_per_kv_head = total_num_heads // total_num_kv_heads
    params_dict = dict(self.named_parameters(remove_duplicate=False))
    loaded_params: set[str] = set()
    for name, loaded_weight in weights:
        # 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]
        if "query_key_value" in name:
            output_dim = getattr(param, "output_dim", None)
            loaded_weight_shape = loaded_weight.shape
            if output_dim is not None:
                loaded_weight = loaded_weight.view(
                    loaded_weight_shape[:output_dim] +
                    (total_num_kv_heads, num_query_heads_per_kv_head + 2,
                     -1) + loaded_weight_shape[output_dim + 1:])
                wq = loaded_weight.narrow(
                    output_dim + 1, 0,
                    num_query_heads_per_kv_head).reshape(
                        *loaded_weight_shape[:output_dim], -1,
                        *loaded_weight_shape[output_dim + 1:])
                wk = loaded_weight.narrow(
                    output_dim + 1, num_query_heads_per_kv_head,
                    1).reshape(*loaded_weight_shape[:output_dim], -1,
                               *loaded_weight_shape[output_dim + 1:])
                wv = loaded_weight.narrow(
                    output_dim + 1, num_query_heads_per_kv_head + 1,
                    1).reshape(*loaded_weight_shape[:output_dim], -1,
                               *loaded_weight_shape[output_dim + 1:])
                loaded_weight = torch.cat([wq, wk, wv], dim=output_dim)

        weight_loader = getattr(param, "weight_loader",
                                default_weight_loader)
        weight_loader(param, loaded_weight)
        loaded_params.add(name)
    return loaded_params

_get_alibi_slopes

_get_alibi_slopes(total_num_heads: int) -> Tensor

Source code in vllm/model_executor/models/falcon.py

def _get_alibi_slopes(total_num_heads: int) -> torch.Tensor:
    closest_power_of_2 = 2**math.floor(math.log2(total_num_heads))
    base = torch.tensor(2**(-(2**-(math.log2(closest_power_of_2) - 3))),
                        dtype=torch.float32)
    powers = torch.arange(1, 1 + closest_power_of_2, dtype=torch.int32)
    slopes = torch.pow(base, powers)

    if closest_power_of_2 != total_num_heads:
        extra_base = torch.tensor(
            2**(-(2**-(math.log2(2 * closest_power_of_2) - 3))),
            dtype=torch.float32)
        num_remaining_heads = min(closest_power_of_2,
                                  total_num_heads - closest_power_of_2)
        extra_powers = torch.arange(1,
                                    1 + 2 * num_remaining_heads,
                                    2,
                                    dtype=torch.int32)
        slopes = torch.cat(
            [slopes, torch.pow(extra_base, extra_powers)], dim=0)

    return slopes