image-match/python/py/Lib/site-packages/transformers/models/olmo/modular_olmo.py

# Copyright 2024 HuggingFace Inc. team. All rights reserved.
#
# This code is based on EleutherAI's GPT-NeoX library and the GPT-NeoX
# and OPT implementations in this library. It has been modified from its
# original forms to accommodate minor architectural differences compared
# to GPT-NeoX and OPT used by the Meta AI team that trained the model.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
#     http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.

from collections.abc import Callable

import torch
import torch.nn as nn
import torch.nn.functional as F

from ...cache_utils import Cache
from ...modeling_rope_utils import dynamic_rope_update
from ...modeling_utils import ALL_ATTENTION_FUNCTIONS
from ...utils import logging
from ...utils.generic import maybe_autocast
from ..llama.modeling_llama import (
    LlamaAttention,
    LlamaDecoderLayer,
    LlamaForCausalLM,
    LlamaMLP,
    LlamaModel,
    LlamaRotaryEmbedding,
    eager_attention_forward,
    rotate_half,
)
from .configuration_olmo import OlmoConfig


logger = logging.get_logger(__name__)


class OlmoLayerNorm(nn.Module):
    """LayerNorm but with no learnable weight or bias."""

    def __init__(self, hidden_size: int) -> None:
        super().__init__()
        self.normalized_shape = (hidden_size,)

    def forward(self, hidden_states: torch.Tensor) -> torch.Tensor:
        orig_dtype = hidden_states.dtype
        return F.layer_norm(hidden_states.to(dtype=torch.float32), self.normalized_shape, None, None, eps=1e-5).to(
            orig_dtype
        )


class OlmoMLP(LlamaMLP):
    def __init__(self, config):
        super().__init__(config)
        self.gate_proj = nn.Linear(self.hidden_size, self.intermediate_size, bias=False)
        self.up_proj = nn.Linear(self.hidden_size, self.intermediate_size, bias=False)
        self.down_proj = nn.Linear(self.intermediate_size, self.hidden_size, bias=False)


# This is identical to LlamaRotaryEmbedding except the output cos and sin are returned
# as float32 rather than the input type.
class OlmoRotaryEmbedding(LlamaRotaryEmbedding):
    @torch.no_grad()
    @dynamic_rope_update  # power user: used with advanced RoPE types (e.g. dynamic rope)
    def forward(self, x, position_ids):
        inv_freq_expanded = self.inv_freq[None, :, None].float().expand(position_ids.shape[0], -1, 1).to(x.device)
        position_ids_expanded = position_ids[:, None, :].float()

        device_type = x.device.type if isinstance(x.device.type, str) and x.device.type != "mps" else "cpu"
        with maybe_autocast(device_type=device_type, enabled=False):  # Force float32
            freqs = (inv_freq_expanded.float() @ position_ids_expanded.float()).transpose(1, 2)
            emb = torch.cat((freqs, freqs), dim=-1)
            cos = emb.cos() * self.attention_scaling
            sin = emb.sin() * self.attention_scaling
        return cos, sin


def apply_rotary_pos_emb(q, k, cos, sin, unsqueeze_dim=1):
    """Applies Rotary Position Embedding to the query and key tensors.

    Args:
        q (`torch.Tensor`): The query tensor.
        k (`torch.Tensor`): The key tensor.
        cos (`torch.Tensor`): The cosine part of the rotary embedding.
        sin (`torch.Tensor`): The sine part of the rotary embedding.
        unsqueeze_dim (`int`, *optional*, defaults to 1):
            The 'unsqueeze_dim' argument specifies the dimension along which to unsqueeze cos[position_ids] and
            sin[position_ids] so that they can be properly broadcasted to the dimensions of q and k. For example, note
            that cos[position_ids] and sin[position_ids] have the shape [batch_size, seq_len, head_dim]. Then, if q and
            k have the shape [batch_size, heads, seq_len, head_dim], then setting unsqueeze_dim=1 makes
            cos[position_ids] and sin[position_ids] broadcastable to the shapes of q and k. Similarly, if q and k have
            the shape [batch_size, seq_len, heads, head_dim], then set unsqueeze_dim=2.
    Returns:
        `tuple(torch.Tensor)` comprising of the query and key tensors rotated using the Rotary Position Embedding.
    """
    q_type, k_type = q.dtype, k.dtype
    cos = cos.unsqueeze(unsqueeze_dim)
    sin = sin.unsqueeze(unsqueeze_dim)
    q_embed = (q * cos) + (rotate_half(q) * sin)
    k_embed = (k * cos) + (rotate_half(k) * sin)
    return q_embed.to(q_type), k_embed.to(k_type)


class OlmoAttention(LlamaAttention):
    def forward(
        self,
        hidden_states: torch.Tensor,
        position_embeddings: tuple[torch.Tensor, torch.Tensor],
        attention_mask: torch.Tensor | None,
        past_key_values: Cache | None = None,
        **kwargs,
    ) -> tuple[torch.Tensor, torch.Tensor | None]:
        input_shape = hidden_states.shape[:-1]
        hidden_shape = (*input_shape, -1, self.head_dim)

        query_states = self.q_proj(hidden_states)
        key_states = self.k_proj(hidden_states)
        value_states = self.v_proj(hidden_states)

        if self.config.clip_qkv is not None:
            query_states.clamp_(min=-self.config.clip_qkv, max=self.config.clip_qkv)
            key_states.clamp_(min=-self.config.clip_qkv, max=self.config.clip_qkv)
            value_states.clamp_(min=-self.config.clip_qkv, max=self.config.clip_qkv)

        query_states = query_states.view(hidden_shape).transpose(1, 2)
        key_states = key_states.view(hidden_shape).transpose(1, 2)
        value_states = value_states.view(hidden_shape).transpose(1, 2)

        cos, sin = position_embeddings
        query_states, key_states = apply_rotary_pos_emb(query_states, key_states, cos, sin)

        if past_key_values is not None:
            key_states, value_states = past_key_values.update(key_states, value_states, self.layer_idx)

        attention_interface: Callable = ALL_ATTENTION_FUNCTIONS.get_interface(
            self.config._attn_implementation, eager_attention_forward
        )

        attn_output, attn_weights = attention_interface(
            self,
            query_states,
            key_states,
            value_states,
            attention_mask,
            dropout=0.0 if not self.training else self.attention_dropout,
            scaling=self.scaling,
            **kwargs,
        )

        attn_output = attn_output.reshape(*input_shape, -1).contiguous()
        attn_output = self.o_proj(attn_output)
        return attn_output, attn_weights


class OlmoDecoderLayer(LlamaDecoderLayer):
    def __init__(self, config: OlmoConfig, layer_idx: int):
        super().__init__(config, layer_idx)
        self.input_layernorm = OlmoLayerNorm(config.hidden_size)
        self.post_attention_layernorm = OlmoLayerNorm(config.hidden_size)
        self.self_attn = OlmoAttention(config=config, layer_idx=layer_idx)


class OlmoModel(LlamaModel):
    def __init__(self, config: OlmoConfig):
        super().__init__(config)
        self.layers = nn.ModuleList(
            [OlmoDecoderLayer(config, layer_idx) for layer_idx in range(config.num_hidden_layers)]
        )
        self.norm = OlmoLayerNorm(config.hidden_size)


class OlmoForCausalLM(LlamaForCausalLM):
    pass


__all__ = [
    "OlmoForCausalLM",
    "OlmoModel",
    "OlmoPreTrainedModel",  # noqa: F822
]