first

README.md

@@ -0,0 +1,157 @@
+---
+tags:
+- albert
+- long context
+language:
+- en
+pipeline_tag: fill-mask
+---
+
+# LSG model 
+**Transformers >= 4.18.0**\
+**This model relies on a custom modeling file, you need to add trust_remote_code=True**\
+**See [\#13467](https://github.com/huggingface/transformers/pull/13467)**
+
+* [Usage](#usage)
+* [Parameters](#parameters)
+* [Sparse selection type](#sparse-selection-type)
+* [Tasks](#tasks)
+
+This model is adapted from [AlBERT-base-v2](https://huggingface.co/albert-base-v2) without additional pretraining. It uses the same number of parameters/layers and the same tokenizer.
+
+
+This model can handle long sequences but faster and more efficiently than Longformer (LED) or BigBird (Pegasus) from the hub and relies on Local + Sparse + Global attention (LSG).
+
+The model requires sequences whose length is a multiple of the block size. The model is "adaptive" and automatically pads the sequences if needed (adaptive=True in config). It is however recommended, thanks to the tokenizer, to truncate the inputs (truncation=True) and optionally to pad with a multiple of the block size (pad_to_multiple_of=...). \
+
+Implemented in PyTorch.
+
+![attn](attn.png)
+
+## Usage
+The model relies on a custom modeling file, you need to add trust_remote_code=True to use it.
+
+```python: 
+from transformers import AutoModel, AutoTokenizer
+
+model = AutoModel.from_pretrained("ccdv/lsg-albert-base-v2-4096", trust_remote_code=True)
+tokenizer = AutoTokenizer.from_pretrained("ccdv/lsg-albert-base-v2-4096")
+``` 
+
+## Parameters
+You can change various parameters like : 
+* the number of global tokens (num_global_tokens=1)
+* local block size (block_size=128)
+* sparse block size (sparse_block_size=128)
+* sparsity factor (sparsity_factor=2)
+* mask_first_token (mask first token since it is redundant with the first global token)
+* see config.json file
+
+Default parameters work well in practice. If you are short on memory, reduce block sizes, increase sparsity factor and remove dropout in the attention score matrix.
+
+```python:
+from transformers import AutoModel
+
+model = AutoModel.from_pretrained("ccdv/lsg-albert-base-v2-4096", 
+    trust_remote_code=True, 
+    num_global_tokens=16,
+    block_size=64,
+    sparse_block_size=64,
+    attention_probs_dropout_prob=0.0
+    sparsity_factor=4,
+    sparsity_type="none",
+    mask_first_token=True
+)
+``` 
+
+## Sparse selection type
+
+There are 5 different sparse selection patterns. The best type is task dependent. \
+Note that for sequences with length < 2*block_size, the type has no effect.
+
+* sparsity_type="norm", select highest norm tokens
+    * Works best for a small sparsity_factor (2 to 4)
+    * Additional parameters:
+        * None
+* sparsity_type="pooling", use average pooling to merge tokens
+    * Works best for a small sparsity_factor (2 to 4)
+    * Additional parameters:
+        * None
+* sparsity_type="lsh", use the LSH algorithm to cluster similar tokens
+    * Works best for a large sparsity_factor (4+)
+    * LSH relies on random projections, thus inference may differ slightly with different seeds
+    * Additional parameters:
+        * lsg_num_pre_rounds=1, pre merge tokens n times before computing centroids
+* sparsity_type="stride", use a striding mecanism per head
+    * Each head will use different tokens strided by sparsify_factor
+    * Not recommended if sparsify_factor > num_heads
+* sparsity_type="block_stride", use a striding mecanism per head
+    * Each head will use block of tokens strided by sparsify_factor
+    * Not recommended if sparsify_factor > num_heads
+
+## Tasks
+Seq2Seq example for summarization:
+```python:
+from transformers import AutoModelForSeq2SeqLM, AutoTokenizer
+
+model = AutoModelForSeq2SeqLM.from_pretrained("ccdv/lsg-albert-base-v2-4096", 
+    trust_remote_code=True, 
+    pass_global_tokens_to_decoder=True, # Pass encoder global tokens to decoder
+)
+tokenizer = AutoTokenizer.from_pretrained("ccdv/lsg-albert-base-v2-4096")
+
+SENTENCE = "This is a test sequence to test the model. " * 300
+token_ids = tokenizer(
+    SENTENCE, 
+    return_tensors="pt", 
+    padding="max_length", # Optional but recommended
+    truncation=True # Optional but recommended
+    )
+output = model(**token_ids)
+```
+
+
+Classification example:
+```python:
+from transformers import AutoModelForSequenceClassification, AutoTokenizer
+
+model = AutoModelForSequenceClassification.from_pretrained("ccdv/lsg-albert-base-v2-4096", 
+    trust_remote_code=True, 
+    pass_global_tokens_to_decoder=True, # Pass encoder global tokens to decoder
+)
+tokenizer = AutoTokenizer.from_pretrained("ccdv/lsg-albert-base-v2-4096")
+
+SENTENCE = "This is a test sequence to test the model. " * 300
+token_ids = tokenizer(
+    SENTENCE, 
+    return_tensors="pt", 
+    #pad_to_multiple_of=... # Optional
+    truncation=True
+    )
+output = model(**token_ids)
+
+> SequenceClassifierOutput(loss=None, logits=tensor([[-0.3051, -0.1762]], grad_fn=<AddmmBackward>), hidden_states=None, attentions=None)
+```
+
+**AlBERT**
+```
+@article{DBLP:journals/corr/abs-1909-11942,
+  author    = {Zhenzhong Lan and
+               Mingda Chen and
+               Sebastian Goodman and
+               Kevin Gimpel and
+               Piyush Sharma and
+               Radu Soricut},
+  title     = {{ALBERT:} {A} Lite {BERT} for Self-supervised Learning of Language
+               Representations},
+  journal   = {CoRR},
+  volume    = {abs/1909.11942},
+  year      = {2019},
+  url       = {http://arxiv.org/abs/1909.11942},
+  archivePrefix = {arXiv},
+  eprint    = {1909.11942},
+  timestamp = {Fri, 27 Sep 2019 13:04:21 +0200},
+  biburl    = {https://dblp.org/rec/journals/corr/abs-1909-11942.bib},
+  bibsource = {dblp computer science bibliography, https://dblp.org}
+}
+```

config.json

@@ -0,0 +1,53 @@
+{
+  "_name_or_path": "roberta-test",
+  "adaptive": true,
+  "architectures": [
+    "LSGAlbertForMaskedLM"
+  ],
+  "attention_probs_dropout_prob": 0,
+  "auto_map": {
+    "AutoConfig": "modeling_lsg_albert.LSGAlbertConfig",
+    "AutoModel": "modeling_lsg_albert.LSGAlbertModel",
+    "AutoModelForMaskedLM": "modeling_lsg_albert.LSGAlbertForMaskedLM",
+    "AutoModelForMultipleChoice": "modeling_lsg_albert.LSGAlbertForMultipleChoice",
+    "AutoModelForPreTraining": "modeling_lsg_albert.LSGAlbertForPreTraining",
+    "AutoModelForQuestionAnswering": "modeling_lsg_albert.LSGAlbertForQuestionAnswering",
+    "AutoModelForSequenceClassification": "modeling_lsg_albert.LSGAlbertForSequenceClassification",
+    "AutoModelForTokenClassification": "modeling_lsg_albert.LSGAlbertForTokenClassification"
+  },
+  "base_model_prefix": "lsg",
+  "block_size": 128,
+  "bos_token_id": 2,
+  "classifier_dropout_prob": 0.1,
+  "down_scale_factor": 1,
+  "embedding_size": 128,
+  "eos_token_id": 3,
+  "gap_size": 0,
+  "hidden_act": "gelu_new",
+  "hidden_dropout_prob": 0,
+  "hidden_size": 768,
+  "initializer_range": 0.02,
+  "inner_group_num": 1,
+  "intermediate_size": 3072,
+  "layer_norm_eps": 1e-12,
+  "lsh_num_pre_rounds": 1,
+  "mask_first_token": true,
+  "max_position_embeddings": 4096,
+  "model_type": "albert",
+  "net_structure_type": 0,
+  "num_attention_heads": 12,
+  "num_global_tokens": 1,
+  "num_hidden_groups": 1,
+  "num_hidden_layers": 12,
+  "num_memory_blocks": 0,
+  "pad_token_id": 0,
+  "pool_with_global": true,
+  "position_embedding_type": "absolute",
+  "sparse_block_size": 128,
+  "sparsity_factor": 2,
+  "sparsity_type": "norm",
+  "torch_dtype": "float32",
+  "transformers_version": "4.20.1",
+  "type_vocab_size": 2,
+  "vocab_size": 30000
+}

modeling_lsg_albert.py

@@ -0,0 +1,1014 @@
+from logging import warn
+from transformers.models.albert.modeling_albert import *
+import torch
+import torch.nn as nn
+from transformers.models.albert.configuration_albert import AlbertConfig
+import sys
+
+AUTO_MAP = {
+        "AutoModel": "modeling_lsg_albert.LSGAlbertModel",
+        "AutoModelForMaskedLM": "modeling_lsg_albert.LSGAlbertForMaskedLM",
+        "AutoModelForPreTraining": "modeling_lsg_albert.LSGAlbertForPreTraining",
+        "AutoModelForMultipleChoice": "modeling_lsg_albert.LSGAlbertForMultipleChoice",
+        "AutoModelForQuestionAnswering": "modeling_lsg_albert.LSGAlbertForQuestionAnswering",
+        "AutoModelForSequenceClassification": "modeling_lsg_albert.LSGAlbertForSequenceClassification",
+        "AutoModelForTokenClassification": "modeling_lsg_albert.LSGAlbertForTokenClassification"
+    }
+
+class LSGAlbertConfig(AlbertConfig):
+    """
+    This class overrides :class:`~transformers.LSGAlbertConfig`. Please check the superclass for the appropriate
+    documentation alongside usage examples.
+    """
+
+    base_model_prefix = "lsg"
+    model_type = "albert"
+
+    def __init__(
+        self,
+        adaptive=True,
+        base_model_prefix="lsg",
+        block_size=128,
+        lsh_num_pre_rounds=1,
+        mask_first_token=False,
+        num_global_tokens=1,
+        pool_with_global=True,
+        sparse_block_size=128,
+        sparsity_factor=2,
+        sparsity_type="norm",
+        **kwargs
+        ):
+        """Constructs LSGAlbertConfig."""
+        super().__init__(**kwargs)
+
+        self.adaptive = adaptive
+        self.auto_map = AUTO_MAP
+        self.base_model_prefix = base_model_prefix
+        self.block_size = block_size
+        self.lsh_num_pre_rounds = lsh_num_pre_rounds
+        self.mask_first_token = mask_first_token
+        self.num_global_tokens = num_global_tokens
+        self.pool_with_global = pool_with_global
+        self.sparse_block_size = sparse_block_size
+        self.sparsity_factor = sparsity_factor
+        self.sparsity_type = sparsity_type
+
+        if sparsity_type not in [None, "none", "norm", "lsh", "pooling", "stride", "block_stride"]:
+            logger.warning(
+                "[WARNING CONFIG]: sparsity_mode not in [None, 'none', 'norm', 'lsh', 'pooling', 'stride', 'block_stride'], setting sparsity_type=None, computation will skip sparse attention")
+            self.sparsity_type = None
+
+        if self.sparsity_type in ["stride", "block_stride"]:
+            if self.sparsity_factor > self.encoder_attention_heads:
+                logger.warning(
+                "[WARNING CONFIG]: sparsity_factor > encoder_attention_heads is not recommended for stride/block_stride sparsity"
+            )
+        
+        if self.num_global_tokens < 1:
+            logger.warning(
+                "[WARNING CONFIG]: num_global_tokens < 1 is not compatible, setting num_global_tokens=1"
+            )
+            self.num_global_tokens = 1
+        elif self.num_global_tokens > 512:
+            logger.warning(
+                "[WARNING CONFIG]: num_global_tokens > 512 is not compatible, setting num_global_tokens=512"
+            )
+            self.num_global_tokens = 512
+        
+        if self.sparsity_factor > 0:
+            assert self.block_size % self.sparsity_factor == 0, "[ERROR CONFIG]: block_size must be divisible by sparsity_factor"
+            assert self.block_size//self.sparsity_factor >= 1, "[ERROR CONFIG]: make sure block_size >= sparsity_factor"
+            
+        
+class BaseSelfAttention(nn.Module):
+    
+    def init_modules(self, config):
+        if config.hidden_size % config.num_attention_heads != 0 and not hasattr(
+            config, "embedding_size"
+        ):
+            raise ValueError(
+                "The hidden size (%d) is not a multiple of the number of attention "
+                "heads (%d)" % (config.hidden_size, config.num_attention_heads)
+            )
+
+        self.num_attention_heads = config.num_attention_heads
+        self.attention_head_size = int(config.hidden_size / config.num_attention_heads)
+        self.all_head_size = self.num_attention_heads * self.attention_head_size
+
+        self.query = nn.Linear(config.hidden_size, self.all_head_size)
+        self.key = nn.Linear(config.hidden_size, self.all_head_size)
+        self.value = nn.Linear(config.hidden_size, self.all_head_size)
+
+        self.dropout = nn.Dropout(config.attention_probs_dropout_prob)
+
+    def transpose_for_scores(self, x):
+        new_x_shape = x.size()[:-1] + (
+            self.num_attention_heads,
+            self.attention_head_size,
+        )
+        x = x.view(*new_x_shape)
+        return x.permute(0, 2, 1, 3)
+
+    def reshape_output(self, context_layer):
+        context_layer = context_layer.permute(0, 2, 1, 3).contiguous()
+        new_context_layer_shape = context_layer.size()[:-2] + (self.all_head_size,)
+        return context_layer.view(*new_context_layer_shape)
+
+    def project_QKV(self, hidden_states):
+
+        query_layer = self.transpose_for_scores(self.query(hidden_states))
+        key_layer = self.transpose_for_scores(self.key(hidden_states))
+        value_layer = self.transpose_for_scores(self.value(hidden_states))
+        return query_layer, key_layer, value_layer
+
+
+class BaseAttentionProduct(nn.Module):
+
+    def __init__(self, config):
+        """
+        Compute attention: softmax(Q @ K.T) @ V
+        """
+        super().__init__()
+        self.dropout = nn.Dropout(config.attention_probs_dropout_prob)
+
+    def forward(self, query_layer, key_layer, value_layer, attention_mask=None):
+        
+        d = query_layer.shape[-1]
+
+        # Take the dot product between "query" and "key" to get the raw attention scores.
+        attention_scores = query_layer @ key_layer.transpose(-1, -2) / math.sqrt(d)
+
+        del query_layer
+        del key_layer
+
+        if attention_mask is not None:
+            # Apply the attention mask is (precomputed for all layers in AlbertModel forward() function)
+            attention_scores = attention_scores + attention_mask
+            del attention_mask
+
+        # Normalize the attention scores to probabilities.
+        attention_probs = nn.Softmax(dim=-1)(attention_scores)
+
+        # This is actually dropping out entire tokens to attend to, which might
+        # seem a bit unusual, but is taken from the original Transformer paper.
+        context_layer = self.dropout(attention_probs) @ value_layer
+
+        return context_layer
+
+
+class CausalAttentionProduct(nn.Module):
+
+    def __init__(self, config):
+        """
+        Compute attention: softmax(Q @ K.T) @ V
+        """
+        super().__init__()
+        self.dropout = nn.Dropout(config.attention_probs_dropout_prob)
+        self.block_size = config.block_size
+
+    def forward(self, query_layer, key_layer, value_layer, attention_mask=None, causal_shape=None):
+        
+        d = query_layer.shape[-1]
+
+        # Take the dot product between "query" and "key" to get the raw attention scores.
+        attention_scores = query_layer @ key_layer.transpose(-1, -2) / math.sqrt(d)
+
+        del query_layer
+        del key_layer
+
+        if attention_mask is not None:
+            # Apply the attention mask is (precomputed for all layers in AlbertModel forward() function)
+            attention_scores = attention_scores + attention_mask
+
+            # Add causal mask
+            causal_shape = (self.block_size, self.block_size) if causal_shape is None else causal_shape
+            causal_mask = torch.tril(
+                torch.ones(*causal_shape, device=attention_mask.device, dtype=attention_scores.dtype), 
+                diagonal=-1
+                ) 
+            causal_mask = causal_mask.T * torch.finfo(attention_scores.dtype).min
+            attention_scores[..., -causal_shape[0]:, -causal_shape[1]:] = causal_mask
+
+            del attention_mask
+
+        # Normalize the attention scores to probabilities.
+        attention_probs = nn.Softmax(dim=-1)(attention_scores)
+
+        # This is actually dropping out entire tokens to attend to, which might
+        # seem a bit unusual, but is taken from the original Transformer paper.
+        context_layer = self.dropout(attention_probs) @ value_layer
+
+        return context_layer
+
+
+class LSGAttentionProduct(nn.Module):
+
+    def __init__(self, config, block_size=None, sparse_block_size=None, sparsity_factor=4, is_causal=False):
+        """
+        Compute block or overlapping blocks attention products
+        """
+        super().__init__()
+ 
+        self.block_size = block_size
+        self.sparse_block_size = sparse_block_size
+        self.sparsity_factor = sparsity_factor
+        self.is_causal = is_causal
+
+        if self.block_size is None:
+            self.block_size = config.block_size
+
+        if self.sparse_block_size is None:
+            self.sparse_block_size = config.sparse_block_size
+
+        # Shape of blocks
+        self.local_shapes = (self.block_size*3, self.block_size)
+        if self.sparse_block_size and self.sparsity_factor > 0:
+            self.sparse_shapes = (self.sparse_block_size*3, self.block_size//self.sparsity_factor)
+
+        if is_causal:
+            self.attention = CausalAttentionProduct(config)
+        else:
+            self.attention = BaseAttentionProduct(config)
+        
+    def build_lsg_inputs(self, hidden_states, sparse_hidden_states, global_hidden_states, is_attn_mask=False):
+        
+        # Build local tokens
+        local_hidden_states = self.reshape_to_local_block(hidden_states, is_attn_mask)
+        del hidden_states
+
+        # Build sparse tokens
+        if sparse_hidden_states is not None:
+            sparse_hidden_states = self.reshape_to_sparse_block(sparse_hidden_states, is_attn_mask)
+        
+        return self.cat_global_sparse_local_tokens(global_hidden_states, sparse_hidden_states, local_hidden_states)
+
+    def forward(
+        self, 
+        query_layer, 
+        key_layer, 
+        value_layer, 
+        attention_mask=None, 
+        sparse_key=None,
+        sparse_value=None, 
+        sparse_mask=None, 
+        global_key=None, 
+        global_value=None, 
+        global_mask=None
+        ):
+
+        # Input batch, heads, length, hidden_size
+        n, h, t, d = query_layer.size()
+        n_blocks = t // self.block_size
+        assert t % self.block_size == 0
+
+        key_layer = self.build_lsg_inputs(
+            key_layer, 
+            sparse_key, 
+            global_key
+            )
+        del sparse_key
+        del global_key
+
+        value_layer = self.build_lsg_inputs(
+            value_layer, 
+            sparse_value, 
+            global_value
+            )
+        del sparse_value
+        del global_value
+
+        attention_mask = self.build_lsg_inputs(
+            attention_mask, 
+            sparse_mask, 
+            global_mask.transpose(-1, -2), 
+            is_attn_mask=True
+            ).transpose(-1, -2)
+        del sparse_mask
+        del global_mask
+
+        # expect (..., t, d) shape
+        # Compute attention
+        context_layer = self.attention(
+                query_layer=self.chunk(query_layer, n_blocks), 
+                key_layer=key_layer,
+                value_layer=value_layer,
+                attention_mask=attention_mask
+                )
+                
+        return context_layer.reshape(n, h, -1, d)
+    
+    def reshape_to_local_block(self, hidden_states, is_attn_mask=False):
+        
+        size, step = self.local_shapes
+        s = (size - step) // 2
+
+        # Pad before block reshaping
+        if is_attn_mask:
+            pad_value = torch.finfo(hidden_states.dtype).min 
+            hidden_states = hidden_states.transpose(-1, -2)
+        else: 
+            pad_value = 0
+
+        hidden_states = torch.nn.functional.pad(
+            hidden_states.transpose(-1, -2), 
+            pad=(s, s),
+            value=pad_value
+            ).transpose(-1, -2)
+
+        # Make blocks
+        hidden_states = hidden_states.unfold(-2, size=size, step=step).transpose(-1, -2)
+
+        # Skip third block if causal
+        if self.is_causal:
+            return hidden_states[..., :size*2//3, :]
+
+        return hidden_states
+
+    def reshape_to_sparse_block(self, hidden_states, is_attn_mask=False):
+        
+        size, step = self.sparse_shapes
+
+        # In case of odd case
+        odd_offset = (step % 2)
+
+        # n, h, t, d*2 + 1
+        size = size*2 
+        s = (size - step) // 2 + odd_offset
+
+        # Pad before block reshaping
+        if is_attn_mask:
+            pad_value = torch.finfo(hidden_states.dtype).min  
+            hidden_states = hidden_states.transpose(-1, -2)
+        else: 
+            pad_value = 0
+
+        hidden_states = torch.nn.functional.pad(
+            hidden_states.transpose(-1, -2), 
+            pad=(s, s),
+            value=pad_value
+            ).transpose(-1, -2)
+
+        # Make blocks
+        hidden_states = hidden_states.unfold(-2, size=size, step=step).transpose(-1, -2)
+
+        # Fix case where block_size == sparsify_factor
+        if odd_offset: 
+            hidden_states = hidden_states[..., :-1, :, :]
+
+        # Indexes for selection
+        u = (size - self.block_size * 3 // self.sparsity_factor) // 2 + odd_offset
+        s = self.sparse_block_size
+
+        # Skip right block if causal
+        if self.is_causal:
+            return hidden_states[..., u-s:u, :]
+
+        u_ = u + odd_offset
+        return torch.cat([hidden_states[..., u-s:u, :], hidden_states[..., -u_:-u_+s, :]], dim=-2)
+
+    def cat_global_sparse_local_tokens(self, x_global, x_sparse=None, x_local=None, dim=-2):
+
+        n, h, b, t, d = x_local.size()
+        x_global = x_global.unsqueeze(-3).expand(-1, -1, b, -1, -1)
+        if x_sparse is not None:
+            return torch.cat([x_global, x_sparse, x_local], dim=dim)
+        return torch.cat([x_global, x_local], dim=dim)
+
+    def chunk(self, x, n_blocks):
+
+        t, d = x.size()[-2:]
+        return x.reshape(*x.size()[:-2], n_blocks, -1, d)
+
+
+class LSGAlbertEmbeddings(AlbertEmbeddings):
+    """
+    Construct the embeddings from word, position and token_type embeddings.
+    """
+
+    def __init__(self, config):
+        super().__init__(config)
+        
+        self.num_global_tokens = config.num_global_tokens
+
+        # Hardcoded but partially trained
+        self.global_embeddings = nn.Embedding(512, embedding_dim=config.embedding_size, )
+
+        self.block_size = config.block_size
+
+    def forward(
+            self,
+            input_ids=None,
+            token_type_ids=None,
+            position_ids=None,
+            inputs_embeds=None,
+            past_key_values_length=0,
+        ) -> torch.Tensor:
+            if input_ids is not None:
+                input_shape = input_ids.size()
+            else:
+                input_shape = inputs_embeds.size()[:-1]
+
+            seq_length = input_shape[1]
+
+            if position_ids is None:
+                position_ids = self.position_ids[:, past_key_values_length : seq_length + past_key_values_length]
+
+            # Setting the token_type_ids to the registered buffer in constructor where it is all zeros, which usually occurs
+            # when its auto-generated, registered buffer helps users when tracing the model without passing token_type_ids, solves
+            # issue #5664
+            if token_type_ids is None:
+                if hasattr(self, "token_type_ids"):
+                    buffered_token_type_ids = self.token_type_ids[:, :seq_length]
+                    buffered_token_type_ids_expanded = buffered_token_type_ids.expand(input_shape[0], seq_length)
+                    token_type_ids = buffered_token_type_ids_expanded
+                else:
+                    token_type_ids = torch.zeros(input_shape, dtype=torch.long, device=self.position_ids.device)
+
+            if inputs_embeds is None:
+                inputs_embeds = self.word_embeddings(input_ids)
+            token_type_embeddings = self.token_type_embeddings(token_type_ids)
+
+            embeddings = inputs_embeds + token_type_embeddings
+            if self.position_embedding_type == "absolute":
+                position_embeddings = self.position_embeddings(position_ids)
+                embeddings += position_embeddings
+
+            n, t, d = embeddings.size()
+        
+            # Add global_tokens
+            indexes = torch.arange(self.num_global_tokens, device=embeddings.device).reshape(1, -1)
+            global_embeddings = self.global_embeddings(indexes) 
+            embeddings = torch.cat([global_embeddings.expand(n, -1, d), embeddings], dim=-2)
+        
+
+            embeddings = self.LayerNorm(embeddings)
+            embeddings = self.dropout(embeddings)
+            return embeddings
+
+
+class LSGAttention(BaseSelfAttention):
+    '''
+    Compute local attention with overlapping blocs
+    Use global attention for tokens with highest norm
+    '''
+    def __init__(self, config):
+        super().__init__()
+
+        self.init_modules(config)
+
+        self.output_dropout = nn.Dropout(config.hidden_dropout_prob)
+        self.dense = nn.Linear(config.hidden_size, config.hidden_size)
+        self.LayerNorm = nn.LayerNorm(config.hidden_size, eps=config.layer_norm_eps)
+
+        self.block_size = config.block_size
+        self.sparse_block_size = config.sparse_block_size
+        self.num_global_tokens = config.num_global_tokens
+        self.sparsity_factor = config.sparsity_factor
+        self.is_causal = config.is_decoder
+        self.is_decoder = config.is_decoder
+
+        self.attention = LSGAttentionProduct(
+            config, 
+            block_size=config.block_size, 
+            sparse_block_size=config.sparse_block_size, 
+            sparsity_factor=self.sparsity_factor, 
+            is_causal=self.is_causal
+            )
+
+        if self.is_causal:
+            self.causal_attention = CausalAttentionProduct(config)
+        self.full_attention = BaseAttentionProduct(config)
+
+        sparse_functions = {
+            "norm": self.get_sparse_tokens_with_norm, 
+            "pooling": self.get_sparse_tokens_with_pooling,
+            "lsh": self.get_sparse_tokens_with_lsh,
+            "stride": self.get_sparse_tokens_with_stride,
+            "block_stride": self.get_sparse_tokens_with_block_stride,
+            }
+        
+        self.sparsity_type = config.sparsity_type
+        self.get_sparse_elements = sparse_functions.get(self.sparsity_type, lambda x, y, z: (None, None, None))
+            
+        if config.sparsity_type == "lsh":
+            self.lsh_num_pre_rounds = config.lsh_num_pre_rounds
+
+    def get_sparse_tokens_with_norm(self, keys, values, mask):
+        
+        if self.sparsity_factor == 1:
+            return keys, values, mask.expand(-1, keys.size()[1], -1, -1)
+
+        with torch.no_grad():
+
+            block_size = min(self.block_size, self.sparse_block_size)
+            key_norm = keys.detach().norm(dim=-1, keepdim=True)
+            key_norm = key_norm * ~mask.transpose(-1, -2).bool()
+            key_norm = self.chunk(key_norm, block_size)
+
+            n, h, b, t, d = key_norm.size()
+            
+            idx = key_norm.argsort(dim=-2) 
+            del key_norm
+            idx += (torch.arange(b, device=keys.device)*t).reshape(1, 1, b, 1, 1)
+
+            split = (t - block_size // self.sparsity_factor, block_size // self.sparsity_factor)
+            sparse_idx = idx.split(split, -2)[-1].reshape(n, h, -1, 1)
+        
+        d = keys.size()[-1]
+        keys = keys.gather(dim=-2, index=sparse_idx.expand(-1, -1, -1, d))
+        values = values.gather(dim=-2, index=sparse_idx.expand(-1, -1, -1, d))
+        mask = mask.expand(-1, h, -1, -1).transpose(-1, -2).gather(dim=-2, index=sparse_idx).transpose(-1, -2)
+
+        return keys, values, mask
+
+    def get_sparse_tokens_with_pooling(self, keys, values, mask):
+        
+        if self.sparsity_factor == 1:
+            return keys, values, mask.expand(-1, keys.size()[1], -1, -1)
+
+        keys = self.chunk(keys, self.sparsity_factor)
+        values = self.chunk(values, self.sparsity_factor)
+
+        n, h, b, t, d = keys.size()
+        mask = mask.reshape(n, 1, b, 1, t)
+        mask = ~mask.transpose(-1, -2).bool()
+
+        keys = keys * mask
+        values = values * mask
+
+        mask = mask.sum(dim=-2)
+        keys = keys.sum(dim=-2) / (mask + 1e-6)
+        values = values.sum(dim=-2) / (mask + 1e-6)
+
+        mask = (1. - mask.clamp(0, 1)) * torch.finfo(mask.dtype).min
+        return keys.reshape(n, h, -1, d), values.reshape(n, h, -1, d), mask.expand(-1, h, -1, -1).transpose(-1, -2)
+
+    def get_sparse_tokens_with_stride(self, keys, values, mask):
+
+        if self.sparsity_factor == 1:
+            return keys, values, mask.expand(-1, keys.size()[1], -1, -1)
+
+        n, h, t, d = keys.size()
+        sparse_idx = torch.arange(t // self.sparsity_factor, device=keys.device) * self.sparsity_factor
+        sparse_idx = sparse_idx.reshape(1, 1, -1, 1) + (torch.arange(h, device=keys.device) % self.sparsity_factor).reshape(1, h, 1, 1)
+        sparse_idx = sparse_idx.expand(n, h, -1, 1)
+
+        keys = keys.gather(dim=-2, index=sparse_idx.expand(-1, -1, -1, d))
+        values = values.gather(dim=-2, index=sparse_idx.expand(-1, -1, -1, d))
+        mask = mask.expand(-1, h, -1, -1).transpose(-1, -2).gather(dim=-2, index=sparse_idx).transpose(-1, -2)
+
+        return keys, values, mask
+
+    def get_sparse_tokens_with_block_stride(self, keys, values, mask):
+
+        if self.sparsity_factor == 1:
+            return keys, values, mask.expand(-1, keys.size()[1], -1, -1)
+
+        n, h, t, d = keys.size()
+
+        t, b = self.block_size, t // self.block_size
+        sparse_idx = torch.arange(t // self.sparsity_factor, device=keys.device)
+        sparse_idx = sparse_idx.reshape(1, 1, 1, -1, 1) + torch.arange(h, device=keys.device).reshape(1, h, 1, 1, 1) * (t // self.sparsity_factor)
+        sparse_idx = (sparse_idx % t) 
+        sparse_idx = sparse_idx + torch.arange(b, device=keys.device).reshape(1, 1, -1, 1, 1) * t
+        sparse_idx = sparse_idx.reshape(1, h, -1, 1).expand(n, h, -1, 1)
+
+        keys = keys.gather(dim=-2, index=sparse_idx.expand(-1, -1, -1, d))
+        values = values.gather(dim=-2, index=sparse_idx.expand(-1, -1, -1, d))
+        mask = mask.expand(-1, h, -1, -1).transpose(-1, -2).gather(dim=-2, index=sparse_idx).transpose(-1, -2)
+
+        return keys, values, mask
+        
+    def get_sparse_tokens_with_lsh(self, keys, values, mask):
+        
+        if self.sparsity_factor == 1:
+            return keys, values, mask.expand(-1, keys.size()[1], -1, -1)
+
+        block_size = min(self.block_size, self.sparse_block_size)
+        keys = self.chunk(keys, block_size)
+        values = self.chunk(values, block_size)
+
+        n, h, b, t, d = keys.size()
+        mask = mask.reshape(n, 1, b, 1, t)
+        mask = ~mask.transpose(-1, -2).bool()
+
+        keys = keys * mask
+        values = values * mask
+        mask = mask.expand(-1, h, -1, -1, -1).float()
+
+        extra_factor = 1
+        
+        for _ in range(self.lsh_num_pre_rounds):
+            keys, values, mask = self.lsh_round(keys, values, mask, t*extra_factor)
+
+        keys, values, mask = self.lsh_round(keys, values, mask, t//self.sparsity_factor)
+        keys /= mask + 1e-8
+        values /= mask + 1e-8
+
+        mask = (1. - mask.clamp(0, 1)) * torch.finfo(mask.dtype).min
+
+        return keys.reshape(n, h, -1, d), values.reshape(n, h, -1, d), mask.transpose(-1, -2).reshape(n, h, 1, -1)
+
+    def lsh_round(self, keys, values, mask, output_size):
+
+        with torch.no_grad():
+
+            n_hashes = output_size // 2
+            n, h, b, t, d = keys.size()
+            binary_mask = mask.clamp(0, 1)
+
+            indexes = (torch.nn.functional.normalize(keys, dim=-1) * binary_mask) @ torch.randn(1, h, 1, d, n_hashes, device=keys.device)
+            indexes = torch.cat([indexes, -indexes], dim=-1).argmax(dim=-1, keepdim=True)
+
+        n, h, b, t, d = keys.size()
+        
+        x_ = torch.zeros(n, h, b, output_size, d, device=keys.device)
+        mask_ = torch.zeros(n, h, b, output_size, 1, device=keys.device)
+        keys = torch.scatter_add(x_, dim=-2, index=indexes.expand(-1, -1, -1, -1, d), src=keys)
+        values = torch.scatter_add(x_, dim=-2, index=indexes.expand(-1, -1, -1, -1, d), src=values)
+        mask = torch.scatter_add(mask_, dim=-2, index=indexes, src=mask)
+
+        return keys[..., :output_size, :], values[..., :output_size, :], mask[..., :output_size, :]
+
+    def forward(
+        self,
+        hidden_states,
+        attention_mask=None,
+        head_mask=None,
+        encoder_hidden_states=None,
+        encoder_attention_mask=None,
+        past_key_value=None,
+        output_attentions=False,
+        ):
+
+        query_layer, key_layer, value_layer = self.project_QKV(hidden_states)
+        outputs = self.not_causal_forward(
+            query_layer,
+            key_layer,
+            value_layer, 
+            attention_mask=attention_mask, 
+            output_attentions=output_attentions
+            )
+        
+        context = outputs[0]
+        context = self.dense(context)
+        context = self.output_dropout(context)
+        context = self.LayerNorm(context + hidden_states)
+
+        outputs = (context, ) + outputs[1:]
+
+        #if head_mask is not None:
+        #    outputs = (outputs[0] * head_mask[:, :, :1, :1], ) + outputs[1:]
+        return outputs
+
+    def not_causal_forward(
+        self,
+        query_layer,
+        key_layer,
+        value_layer,
+        attention_mask=None,
+        output_attentions=False,
+        ):
+
+        n, h, t, d = query_layer.size()
+
+        # Cat global mask
+        attention_mask = torch.nn.functional.pad(attention_mask, (self.num_global_tokens, 0), value=0)
+        
+        # Use normal attention if local attention covers every tokens
+        if t <= 2 * self.block_size + self.num_global_tokens:
+            context_layer = self.full_attention(
+                query_layer=query_layer, 
+                key_layer=key_layer, 
+                value_layer=value_layer, 
+                attention_mask=attention_mask
+                )
+            return (self.reshape_output(context_layer), )
+
+        # Split input into global tokens and other tokens
+        split = (self.num_global_tokens, t - self.num_global_tokens)
+        global_query, query_layer = query_layer.split(split, dim=-2)
+        
+        # Get global_attention
+        bos = self.full_attention(
+            query_layer=global_query, 
+            key_layer=key_layer, 
+            value_layer=value_layer, 
+            attention_mask=attention_mask
+            )
+        
+        # Split K Q M on global and non global
+        global_key, key_layer = key_layer.split(split, dim=-2)
+        global_value, value_layer = value_layer.split(split, dim=-2)
+        global_mask, attention_mask = attention_mask.split(split, dim=-1)
+        
+        n, h, t, d = key_layer.size()
+
+        # Get sparse idx
+        sparse_key, sparse_value, sparse_mask = (None, None, None)
+
+        if self.sparse_block_size and self.sparsity_factor > 0:
+            sparse_key, sparse_value, sparse_mask = self.get_sparse_elements(key_layer, value_layer, attention_mask)
+        
+        # Expand masks on heads
+        attention_mask = attention_mask.expand(-1, h, -1, -1)
+        global_mask = global_mask.expand(-1, h, -1, -1)
+
+        # Compute dot product attention
+        context_layer = self.attention(
+            query_layer, 
+            key_layer, 
+            value_layer, 
+            attention_mask,
+            sparse_key=sparse_key, 
+            sparse_value=sparse_value, 
+            sparse_mask=sparse_mask,
+            global_key=global_key,
+            global_value=global_value,
+            global_mask=global_mask
+            )
+
+        # Merge global and local-sparse tokens
+        context_layer = torch.cat([bos, context_layer], dim=-2)
+        context_layer = self.reshape_output(context_layer)
+        
+        return (context_layer,)
+
+    def chunk(self, x, chunk_size):
+
+        n, h, t, d = x.size()
+        return x.reshape(n, h, -1, chunk_size, d)
+
+
+class LSGAlbertLayer(AlbertLayer):
+
+    def __init__(self, config):
+        super().__init__(config)
+
+        self.attention = LSGAttention(config)
+
+
+class LSGAlbertLayerGroup(AlbertLayerGroup):
+
+    def __init__(self, config):
+        nn.Module.__init__(self)
+
+        self.albert_layers = nn.ModuleList([LSGAlbertLayer(config) for _ in range(config.inner_group_num)])
+
+
+class LSGAlbertTransformer(AlbertTransformer):
+
+    def __init__(self, config):
+        nn.Module.__init__(self)
+
+        self.config = config
+        self.embedding_hidden_mapping_in = nn.Linear(config.embedding_size, config.hidden_size)
+        self.albert_layer_groups = nn.ModuleList([LSGAlbertLayerGroup(config) for _ in range(config.num_hidden_groups)])
+
+
+class LSGAlbertPreTrainedModel(PreTrainedModel):
+    """
+    An abstract class to handle weights initialization and a simple interface for downloading and loading pretrained
+    models.
+    """
+
+    config_class = LSGAlbertConfig
+    load_tf_weights = load_tf_weights_in_albert
+    base_model_prefix = "albert"
+    _keys_to_ignore_on_load_missing = [r"position_ids"]
+
+    def _init_weights(self, module):
+        """Initialize the weights."""
+        if isinstance(module, nn.Linear):
+            # Slightly different from the TF version which uses truncated_normal for initialization
+            # cf https://github.com/pytorch/pytorch/pull/5617
+            module.weight.data.normal_(mean=0.0, std=self.config.initializer_range)
+            if module.bias is not None:
+                module.bias.data.zero_()
+        elif isinstance(module, nn.Embedding):
+            module.weight.data.normal_(mean=0.0, std=self.config.initializer_range)
+            if module.padding_idx is not None:
+                module.weight.data[module.padding_idx].zero_()
+        elif isinstance(module, nn.LayerNorm):
+            module.bias.data.zero_()
+            module.weight.data.fill_(1.0)
+
+
+class LSGAlbertModel(LSGAlbertPreTrainedModel, AlbertModel):
+
+    config_class = LSGAlbertConfig
+    base_model_prefix = "albert"
+
+    def __init__(self, config, add_pooling_layer=True):
+        AlbertPreTrainedModel.__init__(self, config)
+
+        assert hasattr(config, "num_global_tokens")
+        self.num_global_tokens = config.num_global_tokens
+        self.pad_idx = config.pad_token_id
+
+        assert hasattr(config, "block_size") and hasattr(config, "adaptive")
+        self.block_size = config.block_size
+        self.adaptive = config.adaptive
+        self.mask_first_token = config.mask_first_token
+        self.pool_with_global = config.pool_with_global
+
+        self.config = config
+        self.embeddings = LSGAlbertEmbeddings(config)
+        self.encoder = LSGAlbertTransformer(config)
+        if add_pooling_layer:
+            self.pooler = nn.Linear(config.hidden_size, config.hidden_size)
+            self.pooler_activation = nn.Tanh()
+        else:
+            self.pooler = None
+            self.pooler_activation = None
+
+        # Initialize weights and apply final processing
+        self.post_init()
+
+    def forward(
+        self,
+        input_ids=None,
+        attention_mask=None,
+        token_type_ids=None,
+        position_ids=None,
+        head_mask=None,
+        inputs_embeds=None,
+        output_attentions=None,
+        output_hidden_states=None,
+        return_dict=None,
+        ):
+
+        inputs_ = input_ids if input_ids is not None else inputs_embeds
+        n, t = inputs_.size()[:2]
+
+        if attention_mask is None:
+            attention_mask = torch.ones(n, t, device=inputs_.device, dtype=inputs_.dtype)
+        if self.mask_first_token:
+            attention_mask[:,0] = 0
+            
+        b = self.block_size * 2
+        pad = t % self.block_size
+        
+        # Check if t is multiple of block_size and pad
+        if self.adaptive and t > b and pad > 0:
+            pad_length = self.block_size - pad
+            if input_ids is not None:
+                input_ids = torch.nn.functional.pad(input_ids, (0, pad_length), value=self.pad_idx)
+            else:
+                inputs_embeds = torch.nn.functional.pad(inputs_embeds.transpose(-1, -2), (0, pad_length), value=0.).transpose(-1, -2)
+
+            attention_mask = torch.nn.functional.pad(attention_mask, (0, pad_length), value=0)
+
+            if token_type_ids is not None:
+                token_type_ids = torch.nn.functional.pad(token_type_ids, (0, pad_length), value=0)
+            if position_ids is not None:
+                position_ids = torch.nn.functional.pad(position_ids, (0, pad_length), value=0)
+        
+        n, t_ = attention_mask.size()
+
+        encoder_outputs = super().forward(
+            input_ids=input_ids, 
+            attention_mask=attention_mask, 
+            token_type_ids=token_type_ids, 
+            position_ids=position_ids, 
+            head_mask=head_mask, 
+            inputs_embeds=inputs_embeds, 
+            output_attentions=output_attentions, 
+            output_hidden_states=output_hidden_states, 
+            return_dict=return_dict
+            )
+
+        context = encoder_outputs[0]
+        if self.pool_with_global:
+            context[:, self.num_global_tokens] = context[:, 0]
+        
+        diff = t - t_
+        n, _, d = context.size()
+        context = context[..., self.num_global_tokens:, :]
+
+        # Adapt sequence to initial shape
+        if diff < 0:
+            context = context[:, :t]
+        
+        encoder_outputs.last_hidden_state = context
+        sequence_output = encoder_outputs[0]
+        pooled_output = self.pooler(sequence_output) if self.pooler is not None else None
+
+        if not return_dict:
+            return (sequence_output, pooled_output) + encoder_outputs[1:]
+
+        return BaseModelOutputWithPooling(
+            last_hidden_state=sequence_output,
+            pooler_output=pooled_output,
+            hidden_states=encoder_outputs.hidden_states,
+            attentions=encoder_outputs.attentions,
+        )
+
+
+class LSGAlbertForPreTraining(LSGAlbertPreTrainedModel, AlbertForPreTraining):
+
+    def __init__(self, config):
+
+        LSGAlbertPreTrainedModel.__init__(self, config)
+
+        self.albert = LSGAlbertModel(config)
+        self.predictions = AlbertMLMHead(config)
+        self.sop_classifier = AlbertSOPHead(config)
+
+        # Initialize weights and apply final processing
+        self.post_init()
+
+
+class LSGAlbertForMaskedLM(LSGAlbertPreTrainedModel, AlbertForMaskedLM):
+
+    _keys_to_ignore_on_load_unexpected = [r"pooler"]
+
+    def __init__(self, config):
+        LSGAlbertPreTrainedModel.__init__(self, config)
+
+        self.albert = LSGAlbertModel(config, add_pooling_layer=False)
+        self.predictions = AlbertMLMHead(config)
+
+        # Initialize weights and apply final processing
+        self.post_init()
+
+
+class LSGAlbertForSequenceClassification(LSGAlbertPreTrainedModel, AlbertForSequenceClassification):
+
+    def __init__(self, config):
+
+        LSGAlbertPreTrainedModel.__init__(self, config)
+        self.num_labels = config.num_labels
+        self.config = config
+
+        self.albert = LSGAlbertModel(config)
+        self.dropout = nn.Dropout(config.classifier_dropout_prob)
+        self.classifier = nn.Linear(config.hidden_size, self.config.num_labels)
+
+        # Initialize weights and apply final processing
+        self.post_init()
+
+
+class LSGAlbertForTokenClassification(LSGAlbertPreTrainedModel, AlbertForTokenClassification):
+
+    _keys_to_ignore_on_load_unexpected = [r"pooler"]
+
+    def __init__(self, config):
+
+        LSGAlbertPreTrainedModel.__init__(self, config)
+        self.num_labels = config.num_labels
+
+        self.albert = LSGAlbertModel(config, add_pooling_layer=False)
+        classifier_dropout_prob = (
+            config.classifier_dropout_prob
+            if config.classifier_dropout_prob is not None
+            else config.hidden_dropout_prob
+        )
+        self.dropout = nn.Dropout(classifier_dropout_prob)
+        self.classifier = nn.Linear(config.hidden_size, self.config.num_labels)
+
+        # Initialize weights and apply final processing
+        self.post_init()
+
+
+class LSGAlbertForQuestionAnswering(LSGAlbertPreTrainedModel, AlbertForQuestionAnswering):
+
+    _keys_to_ignore_on_load_unexpected = [r"pooler"]
+
+    def __init__(self, config):
+
+        LSGAlbertPreTrainedModel.__init__(self, config)
+        self.num_labels = config.num_labels
+
+        self.albert = LSGAlbertModel(config, add_pooling_layer=False)
+        self.qa_outputs = nn.Linear(config.hidden_size, config.num_labels)
+
+        # Initialize weights and apply final processing
+        self.post_init()
+
+
+class LSGAlbertForMultipleChoice(LSGAlbertPreTrainedModel, AlbertForMultipleChoice):
+
+    def __init__(self, config):
+
+        LSGAlbertPreTrainedModel.__init__(self, config)
+
+        self.albert = LSGAlbertModel(config)
+        self.dropout = nn.Dropout(config.classifier_dropout_prob)
+        self.classifier = nn.Linear(config.hidden_size, 1)
+
+        # Initialize weights and apply final processing
+        self.post_init()
+
+
+def str_to_class(classname):
+    return getattr(sys.modules[__name__], classname)
+
+# Register model in Auto API
+try:
+    LSGAlbertConfig.register_for_auto_class()
+    for key, value in AUTO_MAP.items():
+        str_to_class(value.split(".")[-1]).register_for_auto_class(key)
+except:
+    warn("AutoRegister isn't available, you'll have to manually copy modeling.py after .save_pretrained(...).")
+    warn("Update to transformers >= 4.17.0 to fix.")

pytorch_model.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2f0633412e0ab789a25bb0e397da111223a8b1bf154815084547d9c6ab50e1e6
+size 47288075

special_tokens_map.json

@@ -0,0 +1,15 @@
+{
+  "bos_token": "[CLS]",
+  "cls_token": "[CLS]",
+  "eos_token": "[SEP]",
+  "mask_token": {
+    "content": "[MASK]",
+    "lstrip": true,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "pad_token": "<pad>",
+  "sep_token": "[SEP]",
+  "unk_token": "<unk>"
+}

spiece.model

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:fefb02b667a6c5c2fe27602d28e5fb3428f66ab89c7d6f388e7c8d44a02d0336
+size 760289

tokenizer_config.json

@@ -0,0 +1,23 @@
+{
+  "bos_token": "[CLS]",
+  "cls_token": "[CLS]",
+  "do_lower_case": true,
+  "eos_token": "[SEP]",
+  "keep_accents": false,
+  "mask_token": {
+    "__type": "AddedToken",
+    "content": "[MASK]",
+    "lstrip": true,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "model_max_length": 4096,
+  "name_or_path": "albert-base-v2",
+  "pad_token": "<pad>",
+  "remove_space": true,
+  "sep_token": "[SEP]",
+  "special_tokens_map_file": null,
+  "tokenizer_class": "AlbertTokenizer",
+  "unk_token": "<unk>"
+}