Update hf_prefixlm_converter.py

hf_prefixlm_converter.py

@@ -1,8 +1,6 @@
 """Converts Huggingface Causal LM to Prefix LM.
-
 Conversion does lightweight surgery on a HuggingFace
 Causal LM to convert it to a Prefix LM.
-
 Prefix LMs accepts a `bidirectional_mask` input in `forward`
 and treat the input prompt as the prefix in `generate`.
 """
@@ -12,29 +10,90 @@ from types import MethodType
 from typing import Any, List, MutableMapping, Optional, Tuple, Union
 import torch
 from transformers.models.bloom.modeling_bloom import BaseModelOutputWithPastAndCrossAttentions, BloomForCausalLM, BloomModel, CausalLMOutputWithCrossAttentions, CrossEntropyLoss
-from transformers.models.bloom.modeling_bloom import _expand_mask as _expand_mask_bloom
-from transformers.models.bloom.modeling_bloom import _make_causal_mask as _make_causal_mask_bloom
+
+#depreciated
+#from transformers.models.bloom.modeling_bloom import _expand_mask as _expand_mask_bloom
+def _expand_mask_bloom(mask: torch.Tensor, tgt_length: int) -> torch.BoolTensor:
+    """
+    Expands attention_mask from `[batch_size, src_length]` to `[batch_size, 1, tgt_length, src_length]`.
+    """
+    batch_size, src_length = mask.shape
+    tgt_length = tgt_length if tgt_length is not None else src_length
+
+    expanded_mask = ~(mask[:, None, None, :].to(torch.bool))
+    return expanded_mask.expand(batch_size, 1, tgt_length, src_length)
+
+#from transformers.models.bloom.modeling_bloom import _make_causal_mask as _make_causal_mask_bloom
+
+def _make_causal_mask_bloom(
+    input_ids_shape: torch.Size, device: torch.device, past_key_values_length: int
+) -> torch.BoolTensor:
+    """
+    Make causal mask used for self-attention.
+    """
+    batch_size, target_length = input_ids_shape
+    mask = torch.empty((target_length, target_length + past_key_values_length), dtype=torch.bool, device=device)
+    # ONNX doesn't support `torch.Tensor.triu` properly, thus we use this workaround
+    seq_ids = torch.arange(target_length, device=device)
+    mask[:, past_key_values_length:] = seq_ids[:, None] < seq_ids[None, :]
+
+    if past_key_values_length > 0:
+        mask[:, :past_key_values_length] = False
+
+    expanded_mask = mask[None, None, :, :].expand(batch_size, 1, target_length, target_length + past_key_values_length)
+    return expanded_mask
+
 from transformers.models.bloom.modeling_bloom import logging
 from transformers.models.gpt2.modeling_gpt2 import GPT2LMHeadModel
 from transformers.models.gpt_neo.modeling_gpt_neo import GPTNeoForCausalLM
 from transformers.models.gpt_neox.modeling_gpt_neox import GPTNeoXForCausalLM
 from transformers.models.gptj.modeling_gptj import GPTJForCausalLM
 from transformers.models.opt.modeling_opt import OPTForCausalLM
-from transformers.models.opt.modeling_opt import _expand_mask as _expand_mask_opt
-from transformers.models.opt.modeling_opt import _make_causal_mask as _make_causal_mask_opt
+#from transformers.models.opt.modeling_opt import _expand_mask as _expand_mask_opt
+#from transformers.models.opt.modeling_opt import _make_causal_mask as _make_causal_mask_opt
+
+def _make_causal_mask_opt(
+    input_ids_shape: torch.Size, dtype: torch.dtype, device: torch.device, past_key_values_length: int = 0
+):
+    """
+    Make causal mask used for bi-directional self-attention.
+    """
+    bsz, tgt_len = input_ids_shape
+    mask = torch.full((tgt_len, tgt_len), torch.finfo(dtype).min, device=device)
+    mask_cond = torch.arange(mask.size(-1), device=device)
+    mask.masked_fill_(mask_cond < (mask_cond + 1).view(mask.size(-1), 1), 0)
+    mask = mask.to(dtype)
+
+    if past_key_values_length > 0:
+        mask = torch.cat([torch.zeros(tgt_len, past_key_values_length, dtype=dtype, device=device), mask], dim=-1)
+    return mask[None, None, :, :].expand(bsz, 1, tgt_len, tgt_len + past_key_values_length)
+
+
+def _expand_mask_opt(mask: torch.Tensor, dtype: torch.dtype, tgt_len: Optional[int] = None):
+    """
+    Expands attention_mask from `[bsz, seq_len]` to `[bsz, 1, tgt_seq_len, src_seq_len]`.
+    """
+    bsz, src_len = mask.size()
+    tgt_len = tgt_len if tgt_len is not None else src_len
+
+    expanded_mask = mask[:, None, None, :].expand(bsz, 1, tgt_len, src_len).to(dtype)
+
+    inverted_mask = 1.0 - expanded_mask
+
+    return inverted_mask.masked_fill(inverted_mask.to(torch.bool), torch.finfo(dtype).min)
+
+
 logger = logging.get_logger(__name__)
 _SUPPORTED_GPT_MODELS = (GPT2LMHeadModel, GPTJForCausalLM, GPTNeoForCausalLM, GPTNeoXForCausalLM)
 CAUSAL_GPT_TYPES = Union[GPT2LMHeadModel, GPTJForCausalLM, GPTNeoForCausalLM, GPTNeoXForCausalLM]
 
 def _convert_gpt_causal_lm_to_prefix_lm(model: CAUSAL_GPT_TYPES) -> CAUSAL_GPT_TYPES:
     """Converts a GPT-style Causal LM to a Prefix LM.
-
     Supported HuggingFace model classes:
         - `GPT2LMHeadModel`
         - `GPTNeoForCausalLM`
         - `GPTNeoXForCausalLM`
         - `GPTJForCausalLM`
-
     See `convert_hf_causal_lm_to_prefix_lm` for more details.
     """
     if hasattr(model, '_prefix_lm_converted'):
@@ -44,7 +103,6 @@ def _convert_gpt_causal_lm_to_prefix_lm(model: CAUSAL_GPT_TYPES) -> CAUSAL_GPT_T
 
     def _get_attn_modules(model: CAUSAL_GPT_TYPES) -> List[torch.nn.Module]:
         """Helper that gets a list of the model's attention modules.
-
         Each module has a `bias` buffer used for causal masking. The Prefix LM
         conversion adds logic to dynamically manipulate these biases to support
         Prefix LM attention masking.
@@ -113,10 +171,8 @@ def _convert_gpt_causal_lm_to_prefix_lm(model: CAUSAL_GPT_TYPES) -> CAUSAL_GPT_T
 
 def _convert_bloom_causal_lm_to_prefix_lm(model: BloomForCausalLM) -> BloomForCausalLM:
     """Converts a BLOOM Causal LM to a Prefix LM.
-
     Supported HuggingFace model classes:
         - `BloomForCausalLM`
-
     See `convert_hf_causal_lm_to_prefix_lm` for more details.
     """
     if hasattr(model, '_prefix_lm_converted'):
@@ -270,10 +326,8 @@ def _convert_bloom_causal_lm_to_prefix_lm(model: BloomForCausalLM) -> BloomForCa
 
 def _convert_opt_causal_lm_to_prefix_lm(model: OPTForCausalLM) -> OPTForCausalLM:
     """Converts an OPT Causal LM to a Prefix LM.
-
     Supported HuggingFace model classes:
         - `OPTForCausalLM`
-
     See `convert_hf_causal_lm_to_prefix_lm` for more details.
     """
     if hasattr(model, '_prefix_lm_converted'):
@@ -339,7 +393,6 @@ CAUSAL_LM_TYPES = Union[GPT2LMHeadModel, GPTJForCausalLM, GPTNeoForCausalLM, GPT
 
 def convert_hf_causal_lm_to_prefix_lm(model: CAUSAL_LM_TYPES) -> CAUSAL_LM_TYPES:
     """Converts a HuggingFace Causal LM to a Prefix LM.
-
     Supported HuggingFace model classes:
         - `GPT2LMHeadModel`
         - `GPTNeoForCausalLM`
@@ -347,49 +400,38 @@ def convert_hf_causal_lm_to_prefix_lm(model: CAUSAL_LM_TYPES) -> CAUSAL_LM_TYPES
         - `GPTJForCausalLM`
         - `BloomForCausalLM`
         - `OPTForCausalLM`
-
     Conversion to a Prefix LM is done by modifying the `forward` method, and possibly also the
     `generate` method and/or select underlying methods depending on the model class.
-
     These changes preserve the model API, but add a new input to `forward`: "bidirectional_mask".
-
     Notes on training:
         To actually train the converted model as a Prefix LM, training batches will need to indicate
         the prefix/target structure by including `bidirectional_mask` as part of the batch inputs.
-
         **This is not a standard input and requires custom layers either within or after your dataloader.**
-
         In addition to adding `bidirectional_mask` to the batch, this custom code should modify `labels`
         such that `batch['labels'][batch['bidirectional_mask'] == 1] == -100`.
         That is, the prefix portion of the sequence should not generate any loss. Loss should only be
         generated by the target portion of the sequence.
-
     Notes on `GPTNeoForCausalLM`:
         To simplify the implementation, "global" and "local" attention layers are handled differently.
         For "global" layers, we handle conversion as described above. For "local" layers, which use a
         causal attention mask within a restricted local window, we do not alter the masking.
-
     Notes on `forward` method conversion:
         After conversion, the `forward` method will handle a new input, `bidirectional_mask`,
         which should be a [batch_size, seq_length] byte tensor, where 1 indicates token positions
         belonging to the prefix (prefix tokens can attend to one another bidirectionally), and
         0 indicates token positions belonging to the target.
-
         The new `forward` method will incorporate `bidirectional_mask` (if supplied) into the existing
         causal mask, call the original `forward` method, and (if the causal mask is a buffer) reset
         the causal masks before returning the result.
-
     Notes on `generate` method conversion:
         After conversion, the `generate` method will have the same signature but will internally
         convert all causal masks to be purely bidirectional, call the original `generate` method, and
         (where appropriate) reset the causal masks before returning the result.
-
         This works thanks to the logic of the HuggingFace `generate` API, which first encodes the token
         "prompt" passed to `generate` (which is treated as the prefix) and then sequentially generates
         each new token. Encodings are cached as generation happens, so all prefix tokens can attend to one
         another (as expected in a Prefix LM) and generated tokens can only attend to prefix tokens and
         previously-generated tokens (also as expected in a Prefix LM).
-
     To preserve the API, the original methods are renamed to `_original_forward` and
     `_original_generate`, and replaced with new `forward` and `generate` methods that wrap
     them, respectively. Although implementation details vary by model class.
@@ -405,7 +447,6 @@ def convert_hf_causal_lm_to_prefix_lm(model: CAUSAL_LM_TYPES) -> CAUSAL_LM_TYPES
 
 def add_bidirectional_mask_if_missing(batch: MutableMapping):
     """Attempts to add bidirectional_mask to batch if missing.
-
     Raises:
         KeyError if bidirectional_mask is missing and can't be inferred
     """