Witllm/chatglm/modeling_chatglm.py

import collections
import math
import copy
import os
import gc
import json
import hashlib

import torch
import torch.utils.checkpoint
import torch.nn.functional as F
from torch import nn
from torch.nn.utils import skip_init
from typing import Optional, Tuple, Union, List, Dict, Any

from tqdm import auto as tqdm_lib
from safetensors.torch import storage_ptr, storage_size

from transformers.configuration_utils import PretrainedConfig
from transformers.generation import GenerationConfig

from configuration_chatglm import ChatGLMConfig
from tools import show


class RotaryEmbedding(nn.Module):
    def __init__(self, dim: int, original_impl=False, device=None, dtype=None):
        super().__init__()
        inv_freq = 1.0 / (
            10000 ** (torch.arange(0, dim, 2, device=device).to(dtype=dtype) / dim)
        )
        self.register_buffer("inv_freq", inv_freq)
        self.dim = dim
        self.original_impl = original_impl

    def forward(self, max_seq_len: int, base: int = 10000):
        dtype = self.inv_freq.dtype
        device = self.inv_freq.device
        # $\Theta = {\theta_i = 10000^{\frac{2(i-1)}{d}}, i \in [1, 2, ..., \frac{d}{2}]}$
        theta = 1.0 / (
            base
            ** (
                torch.arange(0, self.dim, 2, dtype=torch.float, device=device)
                / self.dim
            )
        )
        # Create position indexes `[0, 1, ..., max_seq_len - 1]`
        seq_idx = torch.arange(max_seq_len, dtype=torch.float, device=device)
        # Calculate the product of position index and $\theta_i$
        idx_theta = torch.outer(seq_idx, theta).float()
        cache = torch.stack([torch.cos(idx_theta), torch.sin(idx_theta)], dim=-1)
        # this is to mimic the behaviour of complex32, else we will get different results
        if dtype in (torch.float16, torch.bfloat16, torch.int8):
            cache = cache.bfloat16() if dtype == torch.bfloat16 else cache.half()
        return cache


class RMSNorm(torch.nn.Module):
    def __init__(self, normalized_shape, eps=1e-5, device=None, dtype=None, **kwargs):
        super().__init__()
        self.weight = torch.nn.Parameter(
            torch.empty(normalized_shape, device=device, dtype=dtype)
        )
        self.eps = eps

    def forward(self, hidden_states: torch.Tensor):
        input_dtype = hidden_states.dtype
        variance = hidden_states.to(torch.float32).pow(2).mean(-1, keepdim=True)
        hidden_states = hidden_states * torch.rsqrt(variance + self.eps)
        # show.DumpTensorToImage(self.weight, "RMSNorm_weight.png")
        return (self.weight * hidden_states).to(input_dtype)


class CoreAttention(torch.nn.Module):
    def __init__(self, config: ChatGLMConfig, layer_number):
        super(CoreAttention, self).__init__()

        self.attention_softmax_in_fp32 = config.attention_softmax_in_fp32
        self.layer_number = max(1, layer_number)
        projection_size = config.kv_channels * config.num_attention_heads
        # Per attention head and per partition values.
        self.hidden_size_per_partition = projection_size
        self.hidden_size_per_attention_head = (
            projection_size // config.num_attention_heads
        )
        self.num_attention_heads_per_partition = config.num_attention_heads

        coeff = None
        self.norm_factor = math.sqrt(self.hidden_size_per_attention_head)
        coeff = self.layer_number
        self.norm_factor *= coeff
        self.coeff = coeff

        self.attention_dropout = torch.nn.Dropout(config.attention_dropout)

    def forward(self, query_layer, key_layer, value_layer):
        query_layer, key_layer, value_layer = [
            k.permute(1, 2, 0, 3) for k in [query_layer, key_layer, value_layer]
        ]
        if query_layer.shape[2] == key_layer.shape[2]:
            context_layer = torch.nn.functional.scaled_dot_product_attention(
                query_layer, key_layer, value_layer, is_causal=True
            )
        context_layer = context_layer.permute(2, 0, 1, 3)
        new_context_layer_shape = context_layer.size()[:-2] + (
            self.hidden_size_per_partition,
        )
        context_layer = context_layer.reshape(*new_context_layer_shape)
        return context_layer


class SelfAttention(torch.nn.Module):
    def __init__(self, config: ChatGLMConfig, layer_number, device=None):
        super(SelfAttention, self).__init__()
        self.layer_number = max(1, layer_number)
        self.projection_size = config.kv_channels * config.num_attention_heads
        self.hidden_size_per_attention_head = (
            self.projection_size // config.num_attention_heads
        )
        self.num_attention_heads_per_partition = config.num_attention_heads
        self.multi_query_attention = config.multi_query_attention
        self.qkv_hidden_size = 3 * self.projection_size
        self.num_multi_query_groups_per_partition = config.multi_query_group_num
        self.qkv_hidden_size = (
            self.projection_size
            + 2 * self.hidden_size_per_attention_head * config.multi_query_group_num
        )
        self.query_key_value = nn.Linear(
            config.hidden_size,
            self.qkv_hidden_size,
            bias=config.add_bias_linear or config.add_qkv_bias,
            device=device,
            dtype=config.torch_dtype,
        )
        self.core_attention = CoreAttention(config, self.layer_number)
        self.dense = nn.Linear(
            self.projection_size,
            config.hidden_size,
            bias=config.add_bias_linear,
            device=device,
            dtype=config.torch_dtype,
        )

    def apply_rotary_pos_emb(self, x: torch.Tensor, rope: torch.Tensor) -> torch.Tensor:
        # x: [sq, b, np, hn]
        sq, b, np, hn = x.size(0), x.size(1), x.size(2), x.size(3)
        if rope.size(0) != sq:
            raise ("Error rotary_pos_emb size")
        x_rope = x[..., : hn // 2]
        x_pass = x[..., hn // 2 :]
        x_rope = x_rope.reshape(sq, -1, np, hn // 4, 1, 2)
        rope = rope.view(sq, -1, 1, hn // 4, 1, 2)
        roped1 = x_rope[..., 0] * rope[..., 0] - x_rope[..., 1] * rope[..., 1]
        roped2 = x_rope[..., 1] * rope[..., 0] + x_rope[..., 0] * rope[..., 1]
        x_out = torch.cat((roped1, roped2), -1)
        x_out = x_out.flatten(3)
        return torch.cat((x_out, x_pass), dim=-1)

    def forward(self, hidden_states, rotary_pos_emb):
        # hidden_states: [sq, b, h]
        # Attention heads [sq, b, h] --> [sq, b, (np * 3 * hn)]
        mixed_x_layer = self.query_key_value(hidden_states)

        (query_layer, key_layer, value_layer) = mixed_x_layer.split(
            [
                self.num_attention_heads_per_partition
                * self.hidden_size_per_attention_head,
                self.num_multi_query_groups_per_partition
                * self.hidden_size_per_attention_head,
                self.num_multi_query_groups_per_partition
                * self.hidden_size_per_attention_head,
            ],
            dim=-1,
        )
        query_layer = query_layer.view(
            query_layer.size()[:-1]
            + (
                self.num_attention_heads_per_partition,
                self.hidden_size_per_attention_head,
            )
        )
        key_layer = key_layer.view(
            key_layer.size()[:-1]
            + (
                self.num_multi_query_groups_per_partition,
                self.hidden_size_per_attention_head,
            )
        )
        value_layer = value_layer.view(
            value_layer.size()[:-1]
            + (
                self.num_multi_query_groups_per_partition,
                self.hidden_size_per_attention_head,
            )
        )

        # apply relative positional encoding (rotary embedding)
        if rotary_pos_emb is not None:
            query_layer = self.apply_rotary_pos_emb(query_layer, rotary_pos_emb)
            key_layer = self.apply_rotary_pos_emb(key_layer, rotary_pos_emb)

        key_layer = key_layer.unsqueeze(-2)
        key_layer = key_layer.expand(
            -1,
            -1,
            -1,
            self.num_attention_heads_per_partition
            // self.num_multi_query_groups_per_partition,
            -1,
        )
        key_layer = key_layer.contiguous().view(
            key_layer.size()[:2]
            + (
                self.num_attention_heads_per_partition,
                self.hidden_size_per_attention_head,
            )
        )
        value_layer = value_layer.unsqueeze(-2)
        value_layer = value_layer.expand(
            -1,
            -1,
            -1,
            self.num_attention_heads_per_partition
            // self.num_multi_query_groups_per_partition,
            -1,
        )
        value_layer = value_layer.contiguous().view(
            value_layer.size()[:2]
            + (
                self.num_attention_heads_per_partition,
                self.hidden_size_per_attention_head,
            )
        )
        context_layer = self.core_attention(query_layer, key_layer, value_layer)
        output = self.dense(context_layer)  # [sq, b, h]
        return output


class MLP(torch.nn.Module):
    def __init__(self, config: ChatGLMConfig, device=None):
        super(MLP, self).__init__()
        self.add_bias = config.add_bias_linear
        # Project to 4h. If using swiglu double the output width, see https://arxiv.org/pdf/2002.05202.pdf
        self.dense_h_to_4h = nn.Linear(
            config.hidden_size,
            config.ffn_hidden_size * 2,
            bias=self.add_bias,
            device=device,
            dtype=config.torch_dtype,
        )

        def swiglu(x):
            x = torch.chunk(x, 2, dim=-1)
            return F.silu(x[0]) * x[1]

        self.activation_func = swiglu
        self.dense_4h_to_h = nn.Linear(
            config.ffn_hidden_size,
            config.hidden_size,
            bias=self.add_bias,
            device=device,
            dtype=config.torch_dtype,
        )

    def forward(self, hidden_states):
        intermediate_parallel = self.dense_h_to_4h(hidden_states)
        intermediate_parallel = self.activation_func(intermediate_parallel)
        output = self.dense_4h_to_h(intermediate_parallel)
        return output


class GLMBlock(torch.nn.Module):
    """A single transformer layer.

    Transformer layer takes input with size [s, b, h] and returns an
    output of the same size.
    """

    def __init__(self, config: ChatGLMConfig, layer_number, device=None):
        super(GLMBlock, self).__init__()
        self.layer_number = layer_number

        self.apply_residual_connection_post_layernorm = (
            config.apply_residual_connection_post_layernorm
        )

        self.fp32_residual_connection = config.fp32_residual_connection

        LayerNormFunc = RMSNorm
        # Layernorm on the input data.
        self.input_layernorm = LayerNormFunc(
            config.hidden_size,
            eps=config.layernorm_epsilon,
            device=device,
            dtype=config.torch_dtype,
        )
        # Self attention.
        self.self_attention = SelfAttention(config, layer_number, device=device)
        self.hidden_dropout = config.hidden_dropout
        # Layernorm on the attention output
        self.post_attention_layernorm = LayerNormFunc(
            config.hidden_size,
            eps=config.layernorm_epsilon,
            device=device,
            dtype=config.torch_dtype,
        )
        self.mlp = MLP(config, device=device)

    def forward(self, hidden_states, rotary_pos_emb):
        # hidden_states: [s, b, h]
        # Layer norm at the beginning of the transformer layer.
        layernorm_output = self.input_layernorm(hidden_states)
        # Self attention.
        attention_output = self.self_attention(layernorm_output, rotary_pos_emb)
        residual = hidden_states

        layernorm_input = torch.nn.functional.dropout(
            attention_output, p=self.hidden_dropout, training=self.training
        )
        layernorm_input = residual + layernorm_input

        # Layer norm post the self attention.
        layernorm_output = self.post_attention_layernorm(layernorm_input)

        # MLP.
        mlp_output = self.mlp(layernorm_output)

        residual = layernorm_input

        output = torch.nn.functional.dropout(
            mlp_output, p=self.hidden_dropout, training=self.training
        )
        output = residual + output
        return output


class GLMTransformer(torch.nn.Module):
    def __init__(self, config: ChatGLMConfig, device=None):
        super(GLMTransformer, self).__init__()
        self.fp32_residual_connection = config.fp32_residual_connection
        self.post_layer_norm = config.post_layer_norm
        self.num_layers = config.num_layers
        self.layers = []
        for i in range(self.num_layers):
            self.layers.append(GLMBlock(config, i + 1, device=device))
        self.layers = torch.nn.ModuleList(self.layers)
        self.final_layernorm = RMSNorm(
            config.hidden_size,
            eps=config.layernorm_epsilon,
            device=device,
            dtype=config.torch_dtype,
        )

    def forward(self, hidden_states, rotary_pos_emb):
        for index in range(self.num_layers):
            layer = self.layers[index]
            hidden_states = layer(hidden_states, rotary_pos_emb)
        hidden_states = self.final_layernorm(hidden_states)
        return hidden_states


class Embedding(torch.nn.Module):
    def __init__(self, config: ChatGLMConfig, device=None):
        super(Embedding, self).__init__()

        self.hidden_size = config.hidden_size
        self.word_embeddings = nn.Embedding(
            config.padded_vocab_size,
            self.hidden_size,
            dtype=config.torch_dtype,
            device=device,
        )

    def forward(self, input_ids):
        # Embeddings.
        words_embeddings = self.word_embeddings(input_ids)
        embeddings = words_embeddings
        # Data format change to avoid explicit tranposes : [b s h] --> [s b h].
        embeddings = embeddings.transpose(0, 1).contiguous()
        # If the input flag for fp32 residual connection is set, convert for float.
        return embeddings


class ChatGLMModel(nn.Module):
    def __init__(self, config: ChatGLMConfig, device=None, empty_init=True):
        super().__init__()
        init_method = skip_init
        init_kwargs = {}
        if device is not None:
            init_kwargs["device"] = device
        self.embedding = init_method(Embedding, config, **init_kwargs)
        self.num_layers = config.num_layers
        self.multi_query_group_num = config.multi_query_group_num
        self.kv_channels = config.kv_channels
        self.config = config

        # Rotary positional embeddings
        self.seq_length = config.seq_length
        rotary_dim = (
            config.hidden_size // config.num_attention_heads
            if config.kv_channels is None
            else config.kv_channels
        )

        self.rotary_pos_emb = RotaryEmbedding(
            rotary_dim // 2,
            original_impl=config.original_rope,
            device=device,
            dtype=config.torch_dtype,
        )
        self.encoder = init_method(GLMTransformer, config, **init_kwargs)
        self.output_layer = init_method(
            nn.Linear,
            config.hidden_size,
            config.padded_vocab_size,
            bias=False,
            dtype=config.torch_dtype,
            **init_kwargs,
        )

    def forward(
        self,
        input_ids,
        position_ids: Optional[torch.Tensor] = None,
        output_hidden_states: Optional[bool] = None,
        tokenizer=None,
    ):
        output_hidden_states = (
            output_hidden_states
            if output_hidden_states is not None
            else self.config.output_hidden_states
        )
        inputs_embeds = self.embedding(input_ids)

        rotary_pos_emb = self.rotary_pos_emb(self.seq_length)
        # show.DumpTensorToImage(rotary_pos_emb[:, :, 0], "rotary_pos_emb.png", scale=0.1)

        rotary_pos_emb = rotary_pos_emb[position_ids]
        rotary_pos_emb = rotary_pos_emb.transpose(0, 1).contiguous()
        hidden_states_en = self.encoder(inputs_embeds, rotary_pos_emb)
        hidden_states = hidden_states_en[-1:]
        lm_logits = self.output_layer(hidden_states)
        lm_logits = lm_logits.transpose(0, 1).contiguous()

        next_token_logits = lm_logits[:, -1, :]
        probs = nn.functional.softmax(next_token_logits, dim=-1)
        next_tokens = torch.multinomial(probs, num_samples=1).squeeze(1)

        return probs, next_tokens


class ChatGLMForConditionalGeneration(nn.Module):
    def __init__(self, config: ChatGLMConfig, empty_init=True, device=None):
        super().__init__()

        self.max_sequence_length = config.max_length
        self.transformer = ChatGLMModel(config, empty_init=empty_init, device=device)
        self.config = config

        self.main_input_name = "input_ids"
        self.config = config
        self.name_or_path = config.name_or_path
        self.warnings_issued = {}
        self.generation_config = GenerationConfig.from_model_config(config)

    def from_pretrained(
        cls, pretrained_model_name_or_path: Optional[Union[str, os.PathLike]]
    ):
        load_in_8bit = False
        load_in_4bit = False

        pretrained_model_name_or_path = str(pretrained_model_name_or_path)
        resolved_archive_file = os.path.join(
            pretrained_model_name_or_path, "pytorch_model.bin.index.json"
        )
        print(f"loading weights file {resolved_archive_file}")
        with open(resolved_archive_file, "r") as f:
            index = json.loads(f.read())
        shard_filenames = sorted(set(index["weight_map"].values()))
        resolved_archive_file = [
            os.path.join(pretrained_model_name_or_path, f) for f in shard_filenames
        ]
        model = cls._load_pretrained_model(resolved_archive_file)
        model.is_loaded_in_4bit = load_in_4bit
        model.is_loaded_in_8bit = load_in_8bit
        return model

    def _load_state_dict_into_model(self, model_to_load, state_dict, start_prefix):
        metadata = getattr(state_dict, "_metadata", None)
        state_dict = state_dict.copy()
        if metadata is not None:
            state_dict._metadata = metadata
        error_msgs = []

        def load(module: nn.Module, state_dict, prefix=""):
            local_metadata = {} if metadata is None else metadata.get(prefix[:-1], {})
            args = (state_dict, prefix, local_metadata, True, [], [], error_msgs)
            if len([key for key in state_dict if key.startswith(prefix)]) > 0:
                module._load_from_state_dict(*args)

            for name, child in module._modules.items():
                if child is not None:
                    load(child, state_dict, prefix + name + ".")

        load(model_to_load, state_dict, prefix=start_prefix)
        del state_dict
        return error_msgs

    def _load_pretrained_model(cls, resolved_archive_file):
        start_prefix = ""
        model_to_load = cls
        error_msgs = []
        if len(resolved_archive_file) > 1:
            resolved_archive_file = tqdm_lib.tqdm(
                resolved_archive_file, desc="Loading checkpoint shards"
            )
        for shard_file in resolved_archive_file:
            state_dict = torch.load(shard_file, map_location="cpu")

            error_msgs += cls._load_state_dict_into_model(
                model_to_load, state_dict, start_prefix
            )
            del state_dict  # force memory release
            gc.collect()

        print(
            f"All model checkpoint weights were used when initializing {cls.__class__.__name__}.\n"
        )
        return cls

    @torch.inference_mode()
    def chat(
        self,
        tokenizer,
        query: str,
        history: List[Tuple[str, str]] = None,
        role: str = "user",
    ):
        if history is None:
            history = []
        inputs = tokenizer.build_chat_input(query, history=history, role=role)
        inputs = inputs.to(next(self.parameters()).device)

        generation_config = copy.deepcopy(self.generation_config)
        inputs_tensor = inputs["input_ids"]
        input_ids = inputs_tensor.repeat_interleave(
            generation_config.num_return_sequences, dim=0
        )

        outputs = self.sample(
            input_ids,
            generation_config.pad_token_id,
            generation_config.eos_token_id,
            generation_config.output_hidden_states,
            tokenizer,
        )

        outputs = outputs.tolist()[0][:]
        response = tokenizer.decode(outputs)
        history.append({"role": role, "content": query})
        return response, history

    def sample(
        self,
        input_ids: torch.LongTensor,
        pad_token_id: Optional[int] = None,
        eos_token_id: Optional[Union[int, List[int]]] = None,
        output_hidden_states: Optional[bool] = None,
        tokenizer=None,
    ):
        if isinstance(eos_token_id, int):
            eos_token_id = [eos_token_id]
        eos_token_id_tensor = torch.tensor(eos_token_id).to(input_ids.device)

        isFinished = torch.zeros(
            input_ids.shape[0], dtype=torch.long, device=input_ids.device
        )
        # token_count = 0
        while True:
            input_ids_in = input_ids
            batch_size, seq_length = input_ids_in.shape
            position_ids_in = (
                torch.arange(seq_length, dtype=torch.long, device=input_ids.device)
                .unsqueeze(0)
                .repeat(batch_size, 1)
            )
            model_inputs = {"input_ids": input_ids_in, "position_ids": position_ids_in}

            probs, next_tokens = self.transformer(
                **model_inputs,
                output_hidden_states=output_hidden_states,
                tokenizer=tokenizer,
            )

            # finished sentences should add a padding token to next
            pad_token = pad_token_id * isFinished
            next_tokens = next_tokens * (1 - isFinished) + pad_token

            isFinished = isFinished | next_tokens.eq(eos_token_id_tensor)
            if isFinished.min() == 1:  # all batch is finish
                break

            input_ids = torch.cat([input_ids, next_tokens[:, None]], dim=-1)
        return input_ids

    def backward(
        self,
        tokenizer,
        query: str,
    ):
        inputs = tokenizer.build_chat_input(query, history=[], role="user")
        inputs = inputs.to(next(self.parameters()).device)

        generation_config = copy.deepcopy(self.generation_config)
        inputs_tensor = inputs["input_ids"]
        input_ids = inputs_tensor.repeat_interleave(
            generation_config.num_return_sequences, dim=0
        )

        input_ids_in = input_ids
        batch_size, seq_length = input_ids_in.shape
        position_ids_in = (
            torch.arange(seq_length, dtype=torch.long, device=input_ids.device)
            .unsqueeze(0)
            .repeat(batch_size, 1)
        )
        model_inputs = {"input_ids": input_ids_in, "position_ids": position_ids_in}

        probs, next_tokens = self.transformer(
            **model_inputs,
            output_hidden_states=None,
            tokenizer=tokenizer,
        )

        next_tokens = torch.multinomial(probs, num_samples=1).squeeze(1)
        # probs_target = probs
        # probs_target[0, next_tokens] = probs_target[0, next_tokens] * 1.1

        loss = probs[0, next_tokens]
        loss.backward()

        return loss