Update modeling_falcon.py

modeling_falcon.py

@@ -171,6 +171,7 @@ def build_alibi_tensor(attention_mask: torch.Tensor, num_heads: int, dtype: torc
 def dropout_add(x: torch.Tensor, residual: torch.Tensor, prob: float, training: bool) -> torch.Tensor:
     """
     Dropout add function
+
     Args:
         x (`torch.tensor`, *required*):
             input tensor
@@ -223,8 +224,10 @@ class FalconAttention(nn.Module):
     def _split_heads(self, fused_qkv: torch.Tensor) -> Tuple[torch.Tensor, torch.Tensor, torch.Tensor]:
         """
         Split the last dimension into (num_heads, head_dim), results share same memory storage as `fused_qkv`
+
         Args:
             fused_qkv (`torch.tensor`, *required*): [batch_size, seq_length, num_heads * 3 * head_dim]
+
         Returns:
             query: [batch_size, seq_length, num_heads, head_dim] key: [batch_size, seq_length, num_heads, head_dim]
             value: [batch_size, seq_length, num_heads, head_dim]
@@ -253,8 +256,10 @@ class FalconAttention(nn.Module):
     def _merge_heads(self, x: torch.Tensor) -> torch.Tensor:
         """
         Merge heads together over the last dimenstion
+
         Args:
             x (`torch.tensor`, *required*): [batch_size * num_heads, seq_length, head_dim]
+
         Returns:
             torch.tensor: [batch_size, seq_length, num_heads * head_dim]
         """
@@ -485,11 +490,14 @@ class FalconDecoderLayer(nn.Module):
 
 
 FALCON_START_DOCSTRING = r"""
+
     This model inherits from [`PreTrainedModel`]. Check the superclass documentation for the generic methods the
     library implements for all its model (such as downloading or saving, resizing the input embeddings etc.)
+
     This model is also a PyTorch [torch.nn.Module](https://pytorch.org/docs/stable/nn.html#torch.nn.Module) subclass.
     Use it as a regular PyTorch Module and refer to the PyTorch documentation for all matter related to general usage
     and behavior.
+
     Parameters:
         config ([`FalconConfig`]): Model configuration class with all the parameters of the model.
             Initializing with a config file does not load the weights associated with the model, only the
@@ -501,31 +509,40 @@ FALCON_INPUTS_DOCSTRING = r"""
         input_ids (`torch.LongTensor` of shape `(batch_size, input_ids_length)`):
             `input_ids_length` = `sequence_length` if `past_key_values` is `None` else `past_key_values[0][0].shape[2]`
             (`sequence_length` of input past key value states). Indices of input sequence tokens in the vocabulary.
+
             If `past_key_values` is used, only `input_ids` that do not have their past calculated should be passed as
             `input_ids`.
+
             Indices can be obtained using [`AutoTokenizer`]. See [`PreTrainedTokenizer.encode`] and
             [`PreTrainedTokenizer.__call__`] for details.
+
             [What are input IDs?](../glossary#input-ids)
         past_key_values (`Tuple[Tuple[torch.Tensor]]` of length `config.num_hidden_layers`):
             Contains precomputed hidden-states (key and values in the attention blocks) as computed by the model (see
             `past_key_values` output below). Can be used to speed up sequential decoding. The `input_ids` which have
             their past given to this model should not be passed as `input_ids` as they have already been computed.
+
             Each element of `past_key_values` is a tuple (past_key, past_value):
             - past_key: [batch_size * num_heads, head_dim, kv_length]
             - past_value: [batch_size * num_heads, kv_length, head_dim]
         attention_mask (`torch.FloatTensor` of shape `(batch_size, sequence_length)`, *optional*):
             Mask to avoid performing attention on padding token indices. Mask values selected in `[0, 1]`:
+
             - 1 for tokens that are **not masked**,
             - 0 for tokens that are **masked**.
+
             [What are attention masks?](../glossary#attention-mask)
         head_mask (`torch.FloatTensor` of shape `(num_heads,)` or `(num_layers, num_heads)`, *optional*):
             Mask to nullify selected heads of the self-attention modules. Mask values selected in `[0, 1]`:
+
             - 1 indicates the head is **not masked**,
             - 0 indicates the head is **masked**.
+
         inputs_embeds (`torch.FloatTensor` of shape `(batch_size, sequence_length, hidden_size)`, *optional*):
             Optionally, instead of passing `input_ids` you can choose to directly pass an embedded representation. This
             is useful if you want more control over how to convert `input_ids` indices into associated vectors than the
             model's internal embedding lookup matrix.
+
             If `past_key_values` is used, optionally only the last `inputs_embeds` have to be input (see
             `past_key_values`).
         use_cache (`bool`, *optional*):
@@ -622,3 +639,624 @@ class FalconModel(FalconPreTrainedModel):
     def __init__(self, config: FalconConfig):
         super().__init__(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 = nn.Embedding(config.vocab_size, self.embed_dim)
+
+        # Transformer blocks
+        self.h = nn.ModuleList([FalconDecoderLayer(config) for _ in range(config.num_hidden_layers)])
+
+        # Final Layer Norm
+        self.ln_f = LayerNorm(self.embed_dim, eps=config.layer_norm_epsilon)
+
+        self.gradient_checkpointing = False
+
+        # Initialize weights and apply final processing
+        self.post_init()
+
+    def get_input_embeddings(self):
+        return self.word_embeddings
+
+    @staticmethod
+    def _prepare_attn_mask(
+        attention_mask: torch.Tensor, input_shape: Tuple[int, int], past_key_values_length: int
+    ) -> torch.BoolTensor:
+        # Create a causal mask
+        # The attention mask we receive as input should cover the whole extended sequence, including any past
+        # cache, so its shape should be [batch_size, seq_length + past_key_values_length]
+        # The output shape will be [batch_size, 1, seq_length, seq_length + past_key_values_length]
+        if input_shape[1] + past_key_values_length != attention_mask.shape[1]:
+            raise ValueError(
+                "Attention mask shape should be (batch_size, seq_length + past_key_values_length)"
+                f" but is {attention_mask.shape} with input_ids shape {input_shape} and past length"
+                f" {past_key_values_length}."
+            )
+        combined_attention_mask = None
+        device = attention_mask.device
+        _, seq_length = input_shape
+
+        if seq_length > 1:
+            combined_attention_mask = _make_causal_mask(
+                input_shape, device=device, past_key_values_length=past_key_values_length
+            )
+
+        # [batch_size, seq_length + past_key_values_length] -> [batch_size, 1, seq_length, seq_length + past_key_values_length]
+        expanded_attn_mask = _expand_mask(attention_mask, past_key_values_length=past_key_values_length)
+        combined_attention_mask = (
+            expanded_attn_mask if combined_attention_mask is None else expanded_attn_mask | combined_attention_mask
+        )
+
+        return combined_attention_mask
+
+    def set_input_embeddings(self, new_embeddings: torch.Tensor):
+        self.word_embeddings = new_embeddings
+
+    @add_start_docstrings_to_model_forward(FALCON_INPUTS_DOCSTRING)
+    @add_code_sample_docstrings(
+        checkpoint=_CHECKPOINT_FOR_DOC,
+        output_type=BaseModelOutputWithPastAndCrossAttentions,
+        config_class=_CONFIG_FOR_DOC,
+    )
+    def forward(
+        self,
+        input_ids: Optional[torch.LongTensor] = None,
+        past_key_values: Optional[Tuple[Tuple[torch.Tensor, torch.Tensor], ...]] = None,
+        attention_mask: Optional[torch.Tensor] = None,
+        head_mask: Optional[torch.LongTensor] = None,
+        inputs_embeds: Optional[torch.LongTensor] = None,
+        use_cache: Optional[bool] = None,
+        output_attentions: Optional[bool] = None,
+        output_hidden_states: Optional[bool] = None,
+        return_dict: Optional[bool] = None,
+    ) -> Union[Tuple[torch.Tensor, ...], BaseModelOutputWithPastAndCrossAttentions]:
+        output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions
+        output_hidden_states = (
+            output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states
+        )
+        use_cache = use_cache if use_cache is not None else self.config.use_cache
+        return_dict = return_dict if return_dict is not None else self.config.use_return_dict
+
+        if input_ids is not None and inputs_embeds is not None:
+            raise ValueError("You cannot specify both input_ids and inputs_embeds at the same time")
+        elif input_ids is not None:
+            batch_size, seq_length = input_ids.shape
+        elif inputs_embeds is not None:
+            batch_size, seq_length, _ = inputs_embeds.shape
+        else:
+            raise ValueError("You have to specify either input_ids or inputs_embeds")
+
+        if past_key_values is None:
+            past_key_values = tuple([None] * len(self.h))
+        else:
+            past_key_values = self._convert_to_rw_cache(past_key_values)
+
+        # Prepare head mask if needed
+        # 1.0 in head_mask indicate we keep the head
+        # attention_probs has shape batch_size x num_heads x N x N
+        # head_mask has shape n_layer x batch x num_heads x N x N
+        head_mask = self.get_head_mask(head_mask, self.config.num_hidden_layers)
+
+        if inputs_embeds is None:
+            inputs_embeds = self.word_embeddings(input_ids)
+
+        hidden_states = inputs_embeds
+
+        presents = () if use_cache else None
+        all_self_attentions = () if output_attentions else None
+        all_hidden_states = () if output_hidden_states else None
+
+        # Compute alibi tensor: check build_alibi_tensor documentation
+        past_key_values_length = 0
+        if past_key_values[0] is not None:
+            past_key_values_length = past_key_values[0][0].shape[1]  # 1 because RW-cache, not standard format
+        if attention_mask is None:
+            attention_mask = torch.ones((batch_size, seq_length + past_key_values_length), device=hidden_states.device)
+        else:
+            attention_mask = attention_mask.to(hidden_states.device)
+
+        if self.use_alibi:
+            alibi = build_alibi_tensor(attention_mask, self.num_heads, dtype=hidden_states.dtype)
+        else:
+            alibi = None
+
+        causal_mask = self._prepare_attn_mask(
+            attention_mask,
+            input_shape=(batch_size, seq_length),
+            past_key_values_length=past_key_values_length,
+        )
+
+        for i, (block, layer_past) in enumerate(zip(self.h, past_key_values)):
+            if output_hidden_states:
+                all_hidden_states = all_hidden_states + (hidden_states,)
+
+            if self.gradient_checkpointing and self.training:
+                if use_cache:
+                    logger.warning(
+                        "`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`..."
+                    )
+                    use_cache = False
+
+                def create_custom_forward(module):
+                    def custom_forward(*inputs):
+                        # None for past_key_value
+                        return module(*inputs, use_cache=use_cache, output_attentions=output_attentions)
+
+                    return custom_forward
+
+                outputs = torch.utils.checkpoint.checkpoint(
+                    create_custom_forward(block),
+                    hidden_states,
+                    alibi,
+                    causal_mask,
+                    head_mask[i],
+                )
+            else:
+                outputs = block(
+                    hidden_states,
+                    layer_past=layer_past,
+                    attention_mask=causal_mask,
+                    head_mask=head_mask[i],
+                    use_cache=use_cache,
+                    output_attentions=output_attentions,
+                    alibi=alibi,
+                )
+
+            hidden_states = outputs[0]
+            if use_cache is True:
+                presents = presents + (outputs[1],)
+
+            if output_attentions:
+                all_self_attentions = all_self_attentions + (outputs[2 if use_cache else 1],)
+
+        # Add last hidden state
+        hidden_states = self.ln_f(hidden_states)
+
+        if output_hidden_states:
+            all_hidden_states = all_hidden_states + (hidden_states,)
+
+        if presents is not None:
+            presents = self._convert_cache_to_standard_format(presents, batch_size)
+
+        if not return_dict:
+            return tuple(v for v in [hidden_states, presents, all_hidden_states, all_self_attentions] if v is not None)
+
+        return BaseModelOutputWithPastAndCrossAttentions(
+            last_hidden_state=hidden_states,
+            past_key_values=presents,
+            hidden_states=all_hidden_states,
+            attentions=all_self_attentions,
+        )
+
+
+@add_start_docstrings(
+    "The Falcon Model transformer with a language modeling head on top (linear layer with weights tied to the input embeddings).",
+    FALCON_START_DOCSTRING,
+)
+class FalconForCausalLM(FalconPreTrainedModel):
+    _tied_weights_keys = ["lm_head.weight"]
+
+    def __init__(self, config: FalconConfig):
+        super().__init__(config)
+        self.transformer = FalconModel(config)
+        self.lm_head = nn.Linear(config.hidden_size, config.vocab_size, bias=False)
+
+        # Initialize weights and apply final processing
+        self.post_init()
+
+    def get_output_embeddings(self):
+        return self.lm_head
+
+    def set_output_embeddings(self, new_embeddings: torch.Tensor):
+        self.lm_head = new_embeddings
+
+    def prepare_inputs_for_generation(
+        self,
+        input_ids: torch.LongTensor,
+        past_key_values: Optional[torch.Tensor] = None,
+        attention_mask: Optional[torch.Tensor] = None,
+        **kwargs,
+    ) -> dict:
+        if past_key_values is not None:
+            input_ids = input_ids[:, -1:]
+
+        return {
+            "input_ids": input_ids,
+            "past_key_values": past_key_values,
+            "use_cache": kwargs.get("use_cache"),
+            "attention_mask": attention_mask,
+        }
+
+    @add_start_docstrings_to_model_forward(FALCON_INPUTS_DOCSTRING)
+    @add_code_sample_docstrings(
+        checkpoint=_CHECKPOINT_FOR_DOC,
+        output_type=CausalLMOutputWithCrossAttentions,
+        config_class=_CONFIG_FOR_DOC,
+    )
+    def forward(
+        self,
+        input_ids: Optional[torch.LongTensor] = None,
+        past_key_values: Optional[Tuple[Tuple[torch.Tensor, torch.Tensor], ...]] = None,
+        attention_mask: Optional[torch.Tensor] = None,
+        head_mask: Optional[torch.Tensor] = None,
+        inputs_embeds: Optional[torch.Tensor] = None,
+        labels: Optional[torch.Tensor] = None,
+        use_cache: Optional[bool] = None,
+        output_attentions: Optional[bool] = None,
+        output_hidden_states: Optional[bool] = None,
+        return_dict: Optional[bool] = None,
+    ) -> Union[Tuple[torch.Tensor], CausalLMOutputWithCrossAttentions]:
+        r"""
+        labels (`torch.LongTensor` of shape `(batch_size, sequence_length)`, *optional*):
+            Labels for language modeling. Note that the labels **are shifted** inside the model, i.e. you can set
+            `labels = input_ids` Indices are selected in `[-100, 0, ..., config.vocab_size]` All labels set to `-100`
+            are ignored (masked), the loss is only computed for labels in `[0, ..., config.vocab_size]`
+        """
+
+        return_dict = return_dict if return_dict is not None else self.config.use_return_dict
+
+        transformer_outputs = self.transformer(
+            input_ids,
+            past_key_values=past_key_values,
+            attention_mask=attention_mask,
+            head_mask=head_mask,
+            inputs_embeds=inputs_embeds,
+            use_cache=use_cache,
+            output_attentions=output_attentions,
+            output_hidden_states=output_hidden_states,
+            return_dict=return_dict,
+        )
+        hidden_states = transformer_outputs[0]
+
+        lm_logits = self.lm_head(hidden_states)
+
+        loss = None
+        if labels is not None:
+            # Shift so that tokens < n predict n
+            shift_logits = lm_logits[..., :-1, :].contiguous()
+            shift_labels = labels[..., 1:].contiguous()
+            batch_size, seq_length, vocab_size = shift_logits.shape
+            # Flatten the tokens
+            loss_fct = CrossEntropyLoss()
+            loss = loss_fct(
+                shift_logits.view(batch_size * seq_length, vocab_size), shift_labels.view(batch_size * seq_length)
+            )
+
+        if not return_dict:
+            output = (lm_logits,) + transformer_outputs[1:]
+            return ((loss,) + output) if loss is not None else output
+
+        return CausalLMOutputWithCrossAttentions(
+            loss=loss,
+            logits=lm_logits,
+            past_key_values=transformer_outputs.past_key_values,
+            hidden_states=transformer_outputs.hidden_states,
+            attentions=transformer_outputs.attentions,
+        )
+
+    def _reorder_cache(
+        self, past: Tuple[Tuple[torch.Tensor, torch.Tensor], ...], beam_idx: torch.LongTensor
+    ) -> Tuple[Tuple[torch.Tensor, torch.Tensor], ...]:
+        """
+        This function is used to re-order the `past_key_values` cache if [`~PreTrainedModel.beam_search`] or
+        [`~PreTrainedModel.beam_sample`] is called. This is required to match `past_key_values` with the correct
+        beam_idx at every generation step.
+
+        Output shares the same memory storage as `past`.
+        """
+
+        # Get a copy of `beam_idx` on all the devices where we need those indices.
+        device_to_beam_idx = {
+            past_state.device: beam_idx.to(past_state.device) for layer_past in past for past_state in layer_past
+        }
+        reordered_past = tuple(
+            (
+                layer_past[0].index_select(0, device_to_beam_idx[layer_past[0].device]),
+                layer_past[1].index_select(0, device_to_beam_idx[layer_past[0].device]),
+            )
+            for layer_past in past
+        )
+        return reordered_past
+
+
+@add_start_docstrings(
+    """
+    The Falcon Model transformer with a sequence classification head on top (linear layer).
+
+    [`FalconForSequenceClassification`] uses the last token in order to do the classification, as other causal models
+    (e.g. GPT-1) do.
+
+    Since it does classification on the last token, it requires to know the position of the last token. If a
+    `pad_token_id` is defined in the configuration, it finds the last token that is not a padding token in each row. If
+    no `pad_token_id` is defined, it simply takes the last value in each row of the batch. Since it cannot guess the
+    padding tokens when `inputs_embeds` are passed instead of `input_ids`, it does the same (take the last value in
+    each row of the batch).
+    """,
+    FALCON_START_DOCSTRING,
+)
+class FalconForSequenceClassification(FalconPreTrainedModel):
+    def __init__(self, config: FalconConfig):
+        super().__init__(config)
+        self.num_labels = config.num_labels
+        self.transformer = FalconModel(config)
+        self.score = nn.Linear(config.hidden_size, config.num_labels, bias=False)
+
+        # Initialize weights and apply final processing
+        self.post_init()
+
+    @add_start_docstrings_to_model_forward(FALCON_INPUTS_DOCSTRING)
+    @add_code_sample_docstrings(
+        checkpoint=_CHECKPOINT_FOR_DOC,
+        output_type=SequenceClassifierOutputWithPast,
+        config_class=_CONFIG_FOR_DOC,
+    )
+    def forward(
+        self,
+        input_ids: Optional[torch.LongTensor] = None,
+        past_key_values: Optional[Tuple[Tuple[torch.Tensor, torch.Tensor], ...]] = None,
+        attention_mask: Optional[torch.Tensor] = None,
+        head_mask: Optional[torch.Tensor] = None,
+        inputs_embeds: Optional[torch.Tensor] = None,
+        labels: Optional[torch.Tensor] = None,
+        use_cache: Optional[bool] = None,
+        output_attentions: Optional[bool] = None,
+        output_hidden_states: Optional[bool] = None,
+        return_dict: Optional[bool] = None,
+    ) -> Union[Tuple[torch.Tensor], SequenceClassifierOutputWithPast]:
+        r"""
+        labels (`torch.LongTensor` of shape `(batch_size,)`, *optional*):
+            Labels for computing the sequence classification/regression loss. Indices should be in `[0, ...,
+            config.num_labels - 1]`. If `config.num_labels == 1` a regression loss is computed (Mean-Square loss), If
+            `config.num_labels > 1` a classification loss is computed (Cross-Entropy).
+        """
+
+        return_dict = return_dict if return_dict is not None else self.config.use_return_dict
+
+        transformer_outputs = self.transformer(
+            input_ids,
+            past_key_values=past_key_values,
+            attention_mask=attention_mask,
+            head_mask=head_mask,
+            inputs_embeds=inputs_embeds,
+            use_cache=use_cache,
+            output_attentions=output_attentions,
+            output_hidden_states=output_hidden_states,
+            return_dict=return_dict,
+        )
+
+        hidden_states = transformer_outputs[0]
+        logits = self.score(hidden_states)
+
+        if input_ids is not None:
+            batch_size = input_ids.shape[0]
+        else:
+            batch_size = inputs_embeds.shape[0]
+
+        if self.config.pad_token_id is None and batch_size != 1:
+            raise ValueError("Cannot handle batch sizes > 1 if no padding token is defined.")
+        if self.config.pad_token_id is None:
+            sequence_lengths = -1
+        else:
+            if input_ids is not None:
+                sequence_lengths = torch.ne(input_ids, self.config.pad_token_id).sum(dim=-1) - 1
+            else:
+                sequence_lengths = -1
+                logger.warning(
+                    f"{self.__class__.__name__} will not detect padding tokens in `inputs_embeds`. Results may be "
+                    "unexpected if using padding tokens in conjunction with `inputs_embeds.`"
+                )
+
+        pooled_logits = logits[torch.arange(batch_size, device=logits.device), sequence_lengths]
+
+        loss = None
+        if labels is not None:
+            if self.config.problem_type is None:
+                if self.num_labels == 1:
+                    self.config.problem_type = "regression"
+                elif self.num_labels > 1 and (labels.dtype == torch.long or labels.dtype == torch.int):
+                    self.config.problem_type = "single_label_classification"
+                else:
+                    self.config.problem_type = "multi_label_classification"
+
+            if self.config.problem_type == "regression":
+                loss_fct = MSELoss()
+                if self.num_labels == 1:
+                    loss = loss_fct(pooled_logits.squeeze(), labels.squeeze())
+                else:
+                    loss = loss_fct(pooled_logits, labels)
+            elif self.config.problem_type == "single_label_classification":
+                loss_fct = CrossEntropyLoss()
+                loss = loss_fct(pooled_logits, labels)
+            elif self.config.problem_type == "multi_label_classification":
+                loss_fct = BCEWithLogitsLoss()
+                loss = loss_fct(pooled_logits, labels)
+        if not return_dict:
+            output = (pooled_logits,) + transformer_outputs[1:]
+            return ((loss,) + output) if loss is not None else output
+
+        return SequenceClassifierOutputWithPast(
+            loss=loss,
+            logits=pooled_logits,
+            past_key_values=transformer_outputs.past_key_values,
+            hidden_states=transformer_outputs.hidden_states,
+            attentions=transformer_outputs.attentions,
+        )
+
+
+@add_start_docstrings(
+    """
+    Falcon Model with a token classification head on top (a linear layer on top of the hidden-states output) e.g. for
+    Named-Entity-Recognition (NER) tasks.
+    """,
+    FALCON_START_DOCSTRING,
+)
+class FalconForTokenClassification(FalconPreTrainedModel):
+    def __init__(self, config: FalconConfig):
+        super().__init__(config)
+        self.num_labels = config.num_labels
+
+        self.transformer = FalconModel(config)
+        if getattr(config, "classifier_dropout", None) is not None:
+            classifier_dropout = config.classifier_dropout
+        elif getattr(config, "hidden_dropout", None) is not None:
+            classifier_dropout = config.hidden_dropout
+        else:
+            classifier_dropout = 0.1
+        self.dropout = nn.Dropout(classifier_dropout)
+        self.classifier = nn.Linear(config.hidden_size, config.num_labels)
+
+        # Initialize weights and apply final processing
+        self.post_init()
+
+    @add_start_docstrings_to_model_forward(FALCON_INPUTS_DOCSTRING)
+    @add_code_sample_docstrings(
+        checkpoint=_CHECKPOINT_FOR_DOC,
+        output_type=TokenClassifierOutput,
+        config_class=_CONFIG_FOR_DOC,
+    )
+    def forward(
+        self,
+        input_ids: Optional[torch.LongTensor] = None,
+        past_key_values: Optional[Tuple[Tuple[torch.Tensor, torch.Tensor], ...]] = None,
+        attention_mask: Optional[torch.Tensor] = None,
+        head_mask: Optional[torch.Tensor] = None,
+        inputs_embeds: Optional[torch.Tensor] = None,
+        labels: Optional[torch.Tensor] = None,
+        use_cache: Optional[bool] = None,
+        output_attentions: Optional[bool] = None,
+        output_hidden_states: Optional[bool] = None,
+        return_dict: Optional[bool] = None,
+    ) -> Union[Tuple[torch.Tensor], TokenClassifierOutput]:
+        r"""
+        labels (`torch.LongTensor` of shape `(batch_size,)`, *optional*):
+            Labels for computing the sequence classification/regression loss. Indices should be in `[0, ...,
+            config.num_labels - 1]`. If `config.num_labels == 1` a regression loss is computed (Mean-Square loss), If
+            `config.num_labels > 1` a classification loss is computed (Cross-Entropy).
+        """
+
+        return_dict = return_dict if return_dict is not None else self.config.use_return_dict
+
+        transformer_outputs = self.transformer(
+            input_ids,
+            past_key_values=past_key_values,
+            attention_mask=attention_mask,
+            head_mask=head_mask,
+            inputs_embeds=inputs_embeds,
+            use_cache=use_cache,
+            output_attentions=output_attentions,
+            output_hidden_states=output_hidden_states,
+            return_dict=return_dict,
+        )
+
+        hidden_states = transformer_outputs[0]
+        hidden_states = self.dropout(hidden_states)
+        logits = self.classifier(hidden_states)
+
+        loss = None
+        if labels is not None:
+            batch_size, seq_length = labels.shape
+            loss_fct = CrossEntropyLoss()
+            loss = loss_fct(
+                logits.view(batch_size * seq_length, self.num_labels), labels.view(batch_size * seq_length)
+            )
+
+        if not return_dict:
+            output = (logits,) + transformer_outputs[2:]
+            return ((loss,) + output) if loss is not None else output
+
+        return TokenClassifierOutput(
+            loss=loss,
+            logits=logits,
+            hidden_states=transformer_outputs.hidden_states,
+            attentions=transformer_outputs.attentions,
+        )
+
+
+@add_start_docstrings(
+    """
+    The Falcon Model transformer with a span classification head on top for extractive question-answering tasks like
+    SQuAD (a linear layers on top of the hidden-states output to compute `span start logits` and `span end logits`).
+    """,
+    FALCON_START_DOCSTRING,
+)
+class FalconForQuestionAnswering(FalconPreTrainedModel):
+    def __init__(self, config):
+        super().__init__(config)
+        self.transformer = FalconModel(config)
+        self.qa_outputs = nn.Linear(config.hidden_size, 2)
+
+        # Initialize weights and apply final processing
+        self.post_init()
+
+    @add_start_docstrings_to_model_forward(FALCON_INPUTS_DOCSTRING)
+    def forward(
+        self,
+        input_ids: Optional[torch.LongTensor] = None,
+        attention_mask: Optional[torch.FloatTensor] = None,
+        head_mask: Optional[torch.FloatTensor] = None,
+        inputs_embeds: Optional[torch.FloatTensor] = None,
+        start_positions: Optional[torch.LongTensor] = None,
+        end_positions: Optional[torch.LongTensor] = None,
+        output_attentions: Optional[bool] = None,
+        output_hidden_states: Optional[bool] = None,
+        return_dict: Optional[bool] = None,
+    ) -> Union[Tuple, QuestionAnsweringModelOutput]:
+        r"""
+        start_positions (`torch.LongTensor` of shape `(batch_size,)`, *optional*):
+            Labels for position (index) of the start of the labelled span for computing the token classification loss.
+            Positions are clamped to the length of the sequence (`sequence_length`). Position outside of the sequence
+            are not taken into account for computing the loss.
+        end_positions (`torch.LongTensor` of shape `(batch_size,)`, *optional*):
+            Labels for position (index) of the end of the labelled span for computing the token classification loss.
+            Positions are clamped to the length of the sequence (`sequence_length`). Position outside of the sequence
+            are not taken into account for computing the loss.
+        """
+        return_dict = return_dict if return_dict is not None else self.config.use_return_dict
+
+        outputs = self.transformer(
+            input_ids,
+            attention_mask=attention_mask,
+            head_mask=head_mask,
+            inputs_embeds=inputs_embeds,
+            output_attentions=output_attentions,
+            output_hidden_states=output_hidden_states,
+            return_dict=return_dict,
+        )
+
+        sequence_output = outputs[0]
+
+        logits = self.qa_outputs(sequence_output)
+        start_logits, end_logits = logits.split(1, dim=-1)
+        start_logits = start_logits.squeeze(-1).contiguous()
+        end_logits = end_logits.squeeze(-1).contiguous()
+
+        total_loss = None
+        if start_positions is not None and end_positions is not None:
+            # If we are on multi-GPU, split add a dimension
+            if len(start_positions.size()) > 1:
+                start_positions = start_positions.squeeze(-1)
+            if len(end_positions.size()) > 1:
+                end_positions = end_positions.squeeze(-1)
+            # sometimes the start/end positions are outside our model inputs, we ignore these terms
+            ignored_index = start_logits.size(1)
+            start_positions = start_positions.clamp(0, ignored_index)
+            end_positions = end_positions.clamp(0, ignored_index)
+
+            loss_fct = CrossEntropyLoss(ignore_index=ignored_index)
+            start_loss = loss_fct(start_logits, start_positions)
+            end_loss = loss_fct(end_logits, end_positions)
+            total_loss = (start_loss + end_loss) / 2
+
+        if not return_dict:
+            output = (start_logits, end_logits) + outputs[2:]
+            return ((total_loss,) + output) if total_loss is not None else output
+
+        return QuestionAnsweringModelOutput(
+            loss=total_loss,
+            start_logits=start_logits,
+            end_logits=end_logits,
+            hidden_states=outputs.hidden_states,
+            attentions=outputs.attentions,
+        )