Add gemma 2

docs/source/en/index.md

@@ -145,6 +145,7 @@ Flax), PyTorch, and/or TensorFlow.
 |                  [Funnel Transformer](model_doc/funnel)                  |       ✅        |         ✅         |      ❌      |
 |                          [Fuyu](model_doc/fuyu)                          |       ✅        |         ❌         |      ❌      |
 |                         [Gemma](model_doc/gemma)                         |       ✅        |         ❌         |      ✅      |
+|                        [Gemma2](model_doc/gemma2)                        |       ✅        |         ❌         |      ❌      |
 |                           [GIT](model_doc/git)                           |       ✅        |         ❌         |      ❌      |
 |                          [GLPN](model_doc/glpn)                          |       ✅        |         ❌         |      ❌      |
 |                       [GPT Neo](model_doc/gpt_neo)                       |       ✅        |         ❌         |      ✅      |

src/transformers/__init__.py

@@ -435,6 +435,7 @@
     ],
     "models.fuyu": ["FuyuConfig"],
     "models.gemma": ["GemmaConfig"],
+    "models.gemma2": ["Gemma2Config"],
     "models.git": [
         "GitConfig",
         "GitProcessor",
@@ -2181,6 +2182,15 @@
             "GemmaPreTrainedModel",
         ]
     )
+    _import_structure["models.gemma2"].extend(
+        [
+            "Gemma2ForCausalLM",
+            "Gemma2ForSequenceClassification",
+            "Gemma2ForTokenClassification",
+            "Gemma2Model",
+            "Gemma2PreTrainedModel",
+        ]
+    )
     _import_structure["models.git"].extend(
         [
             "GitForCausalLM",
@@ -5062,6 +5072,7 @@
     )
     from .models.fuyu import FuyuConfig
     from .models.gemma import GemmaConfig
+    from .models.gemma2 import Gemma2Config
     from .models.git import (
         GitConfig,
         GitProcessor,
@@ -6694,6 +6705,13 @@
             GemmaModel,
             GemmaPreTrainedModel,
         )
+        from .models.gemma2 import (
+            Gemma2ForCausalLM,
+            Gemma2ForSequenceClassification,
+            Gemma2ForTokenClassification,
+            Gemma2Model,
+            Gemma2PreTrainedModel,
+        )
         from .models.git import (
             GitForCausalLM,
             GitModel,

src/transformers/cache_utils.py

@@ -970,3 +970,125 @@ def get_max_length(self) -> Optional[int]:
         # in theory there is no limit because the sliding window size is fixed
         # no matter how long the sentence is
         return None
+
+
+class HybridCache(Cache):
+    def __init__(self, config: PretrainedConfig, max_batch_size, max_cache_len, device="cpu", dtype=None) -> None:
+        if not hasattr(config, "sliding_window") or config.sliding_window is None:
+            raise ValueError(
+                "Setting `cache_implementation` to 'sliding_window' requires the model config supporting "
+                "sliding window attention, please check if there is a `sliding_window` field in the model "
+                "config and it's not set to None."
+            )
+        self.max_cache_len = max_cache_len
+        self.max_batch_size = max_batch_size
+        # Some model define a custom `head_dim` != config.hidden_size // config.num_attention_heads
+        self.head_dim = (
+            config.head_dim if hasattr(config, "head_dim") else config.hidden_size // config.num_attention_heads
+        )
+
+        self.dtype = dtype if dtype is not None else torch.float32
+        self.num_key_value_heads = (
+            config.num_attention_heads if config.num_key_value_heads is None else config.num_key_value_heads
+        )
+        self.is_sliding = torch.tensor(
+            [i % 2 for i in range(config.num_hidden_layers)], dtype=torch.bool, device=device
+        )
+        self.key_cache: List[torch.Tensor] = []
+        self.value_cache: List[torch.Tensor] = []
+        global_cache_shape = (max_batch_size, self.num_key_value_heads, max_cache_len, self.head_dim)
+        sliding_cache_shape = (
+            max_batch_size,
+            self.num_key_value_heads,
+            min(config.sliding_window, max_cache_len),
+            self.head_dim,
+        )
+        for i in range(config.num_hidden_layers):
+            # Note: `mark_static_address` is used to tag the cache as an fixed data pointer, preventing cuda graph
+            # breaks when updating the cache.
+            cache_shape = global_cache_shape if not self.is_sliding[i] else sliding_cache_shape
+            new_layer_key_cache = torch.zeros(cache_shape, dtype=self.dtype, device=device)
+            new_layer_value_cache = torch.zeros(cache_shape, dtype=self.dtype, device=device)
+            torch._dynamo.mark_static_address(new_layer_key_cache)
+            torch._dynamo.mark_static_address(new_layer_value_cache)
+            self.key_cache.append(new_layer_key_cache)
+            self.value_cache.append(new_layer_value_cache)
+
+    def _sliding_update(self, cache_position, layer_idx, key_states, value_states, k_out, v_out, max_cache_len):
+        if cache_position.shape[0] > max_cache_len:
+            k_out = key_states[:, :, -max_cache_len:, :]
+            v_out = value_states[:, :, -max_cache_len:, :]
+            # Assumption: caches are all zeros at this point, `+=` is equivalent to `=` but compile-friendly
+            self.key_cache[layer_idx] += k_out
+            self.value_cache[layer_idx] += v_out
+            # we should return the whole states instead of k_out, v_out to take the whole prompt
+            # into consideration when building kv cache instead of just throwing away tokens outside of the window
+            return key_states, value_states
+
+        slicing = torch.ones(max_cache_len, dtype=torch.long, device=value_states.device).cumsum(0)
+        cache_position = cache_position.clamp(0, max_cache_len - 1)
+        to_shift = cache_position >= max_cache_len - 1
+        indices = (slicing + to_shift[-1].int() - 1) % max_cache_len
+        k_out = k_out[:, :, indices]
+        v_out = v_out[:, :, indices]
+
+        k_out[:, :, cache_position] = key_states
+        v_out[:, :, cache_position] = value_states
+        # `_.zero()` followed by `+=` is equivalent `=`, but compile-friendly (without graph breaks due to assignment)
+        self.key_cache[layer_idx].zero_()
+        self.value_cache[layer_idx].zero_()
+
+        self.key_cache[layer_idx] += k_out
+        self.value_cache[layer_idx] += v_out
+        return k_out, v_out
+
+    def _static_update(self, cache_position, layer_idx, key_states, value_states, k_out, v_out, max_cache_len):
+        k_out[:, :, cache_position] = key_states
+        v_out[:, :, cache_position] = value_states
+
+        self.key_cache[layer_idx] = k_out
+        self.value_cache[layer_idx] = v_out
+        return k_out, v_out
+
+    def update(
+        self,
+        key_states: torch.Tensor,
+        value_states: torch.Tensor,
+        layer_idx: int,
+        cache_kwargs: Optional[Dict[str, Any]] = None,
+        sliding_window: Optional[int] = None,
+    ) -> Tuple[torch.Tensor]:
+        cache_position = cache_kwargs.get("cache_position")
+        self.key_cache[layer_idx] = self.key_cache[layer_idx].to(device=key_states.device)
+        self.value_cache[layer_idx] = self.value_cache[layer_idx].to(device=value_states.device)
+        k_out = self.key_cache[layer_idx]
+        v_out = self.value_cache[layer_idx]
+        if sliding_window:
+            update_fn = self._sliding_update
+        else:
+            update_fn = self._static_update
+
+        return update_fn(
+            cache_position,
+            layer_idx,
+            key_states,
+            value_states,
+            k_out,
+            v_out,
+            k_out.shape[2],
+        )
+
+    def get_max_length(self) -> Optional[int]:
+        # in theory there is no limit because the sliding window size is fixed
+        # no matter how long the sentence is
+        return self.max_cache_len
+
+    def get_seq_length(self, layer_idx: Optional[int] = 0) -> int:
+        return None
+
+    def reset(self):
+        """Resets the cache values while preserving the objects"""
+        for layer_idx in range(len(self.key_cache)):
+            # In-place ops prevent breaking the static address
+            self.key_cache[layer_idx].zero_()
+            self.value_cache[layer_idx].zero_()

src/transformers/generation/configuration_utils.py

@@ -400,7 +400,7 @@ def __init__(self, **kwargs):
         # Cache implementation
         self.cache_implementation = kwargs.pop("cache_implementation", None)
         self.cache_config = kwargs.pop("cache_config", None)
-        if self.cache_implementation is not None:
+        if self.cache_implementation is not None and self.cache_implementation in NEEDS_CACHE_CONFIG:
             cache_config_class = NEEDS_CACHE_CONFIG[self.cache_implementation]
             if self.cache_config is None:
                 self.cache_config = cache_config_class()

src/transformers/generation/utils.py

@@ -28,6 +28,7 @@
     Cache,
     DynamicCache,
     HQQQuantizedCache,
+    HybridCache,
     QuantizedCacheConfig,
     QuantoQuantizedCache,
     SlidingWindowCache,
@@ -112,7 +113,7 @@
 if is_accelerate_available():
     from accelerate.hooks import AlignDevicesHook, add_hook_to_module
 
-NEED_SETUP_CACHE_CLASSES_MAPPING = {"static": StaticCache, "sliding_window": SlidingWindowCache}
+NEED_SETUP_CACHE_CLASSES_MAPPING = {"static": StaticCache, "sliding_window": SlidingWindowCache, "hybrid": HybridCache}
 QUANT_BACKEND_CLASSES_MAPPING = {"quanto": QuantoQuantizedCache, "HQQ": HQQQuantizedCache}
 
 
@@ -1395,10 +1396,12 @@ def _get_initial_cache_position(self, input_ids, model_kwargs):
 
         past_length = 0
         if model_kwargs.get("past_key_values") is not None:
-            if isinstance(model_kwargs["past_key_values"], Cache):
-                past_length = model_kwargs["past_key_values"].get_seq_length()
-            else:
-                past_length = model_kwargs["past_key_values"][0][0].shape[2]
+            cache = model_kwargs["past_key_values"]
+            if not isinstance(cache, Cache):
+                past_length = cache[0][0].shape[2]
+            elif hasattr(cache, "get_seq_length"):
+                past_length = cache.get_seq_length()
+
         if "inputs_embeds" in model_kwargs:
             cur_len = model_kwargs["inputs_embeds"].shape[1]
         else:
@@ -1739,7 +1742,9 @@ def generate(
                         "issue: https://github.com/huggingface/transformers/issues/28981"
                     )
                 model_kwargs["past_key_values"] = self._get_cache(
-                    generation_config.cache_implementation, batch_size, generation_config.max_length
+                    generation_config.cache_implementation,
+                    getattr(generation_config, "num_beams", 1) * batch_size,
+                    generation_config.max_length,
                 )
             elif generation_config.cache_implementation == "quantized":
                 if not self._supports_quantized_cache:

src/transformers/models/__init__.py

@@ -92,6 +92,7 @@
     funnel,
     fuyu,
     gemma,
+    gemma2,
     git,
     glpn,
     gpt2,

src/transformers/models/auto/configuration_auto.py

@@ -108,6 +108,7 @@
         ("funnel", "FunnelConfig"),
         ("fuyu", "FuyuConfig"),
         ("gemma", "GemmaConfig"),
+        ("gemma2", "Gemma2Config"),
         ("git", "GitConfig"),
         ("glpn", "GLPNConfig"),
         ("gpt-sw3", "GPT2Config"),
@@ -385,6 +386,7 @@
         ("funnel", "Funnel Transformer"),
         ("fuyu", "Fuyu"),
         ("gemma", "Gemma"),
+        ("gemma2", "Gemma2"),
         ("git", "GIT"),
         ("glpn", "GLPN"),
         ("gpt-sw3", "GPT-Sw3"),

src/transformers/models/auto/modeling_auto.py

@@ -105,6 +105,7 @@
         ("fsmt", "FSMTModel"),
         ("funnel", ("FunnelModel", "FunnelBaseModel")),
         ("gemma", "GemmaModel"),
+        ("gemma2", "Gemma2Model"),
         ("git", "GitModel"),
         ("glpn", "GLPNModel"),
         ("gpt-sw3", "GPT2Model"),
@@ -454,6 +455,7 @@
         ("falcon", "FalconForCausalLM"),
         ("fuyu", "FuyuForCausalLM"),
         ("gemma", "GemmaForCausalLM"),
+        ("gemma2", "Gemma2ForCausalLM"),
         ("git", "GitForCausalLM"),
         ("gpt-sw3", "GPT2LMHeadModel"),
         ("gpt2", "GPT2LMHeadModel"),
@@ -863,6 +865,7 @@
         ("fnet", "FNetForSequenceClassification"),
         ("funnel", "FunnelForSequenceClassification"),
         ("gemma", "GemmaForSequenceClassification"),
+        ("gemma2", "Gemma2ForSequenceClassification"),
         ("gpt-sw3", "GPT2ForSequenceClassification"),
         ("gpt2", "GPT2ForSequenceClassification"),
         ("gpt_bigcode", "GPTBigCodeForSequenceClassification"),
@@ -1044,6 +1047,7 @@
         ("fnet", "FNetForTokenClassification"),
         ("funnel", "FunnelForTokenClassification"),
         ("gemma", "GemmaForTokenClassification"),
+        ("gemma2", "Gemma2ForTokenClassification"),
         ("gpt-sw3", "GPT2ForTokenClassification"),
         ("gpt2", "GPT2ForTokenClassification"),
         ("gpt_bigcode", "GPTBigCodeForTokenClassification"),

src/transformers/models/auto/tokenization_auto.py

@@ -188,6 +188,13 @@
                     "GemmaTokenizerFast" if is_tokenizers_available() else None,
                 ),
             ),
+            (
+                "gemma2",
+                (
+                    "GemmaTokenizer" if is_sentencepiece_available() else None,
+                    "GemmaTokenizerFast" if is_tokenizers_available() else None,
+                ),
+            ),
             ("git", ("BertTokenizer", "BertTokenizerFast" if is_tokenizers_available() else None)),
             ("gpt-sw3", ("GPTSw3Tokenizer" if is_sentencepiece_available() else None, None)),
             ("gpt2", ("GPT2Tokenizer", "GPT2TokenizerFast" if is_tokenizers_available() else None)),

src/transformers/models/gemma/diff_gemma.py

@@ -257,6 +257,7 @@ def __init__(self, config: GemmaConfig, layer_idx: Optional[int] = None):
         self.max_position_embeddings = config.max_position_embeddings
         self.rope_theta = config.rope_theta
         self.is_causal = True
+        self.scaling = 1 / math.sqrt(config.head_dim)
 
         if self.hidden_size % self.num_heads != 0:
             raise ValueError(
@@ -305,7 +306,7 @@ def forward(
         key_states = repeat_kv(key_states, self.num_key_value_groups)
         value_states = repeat_kv(value_states, self.num_key_value_groups)
 
-        attn_weights = torch.matmul(query_states, key_states.transpose(2, 3)) / math.sqrt(self.head_dim)
+        attn_weights = torch.matmul(query_states, key_states.transpose(2, 3)) * self.scaling
 
         if attention_mask is not None:  # no matter the length, we just slice it
             causal_mask = attention_mask[:, :, :, : key_states.shape[-2]]

src/transformers/models/gemma/modeling_gemma.py

@@ -240,6 +240,7 @@ def __init__(self, config: GemmaConfig, layer_idx: Optional[int] = None):
         self.max_position_embeddings = config.max_position_embeddings
         self.rope_theta = config.rope_theta
         self.is_causal = True
+        self.scaling = 1 / math.sqrt(config.head_dim)
 
         if self.hidden_size % self.num_heads != 0:
             raise ValueError(
@@ -288,7 +289,7 @@ def forward(
         key_states = repeat_kv(key_states, self.num_key_value_groups)
         value_states = repeat_kv(value_states, self.num_key_value_groups)
 
-        attn_weights = torch.matmul(query_states, key_states.transpose(2, 3)) / math.sqrt(self.head_dim)
+        attn_weights = torch.matmul(query_states, key_states.transpose(2, 3)) * self.scaling
 
         if attention_mask is not None:  # no matter the length, we just slice it
             causal_mask = attention_mask[:, :, :, : key_states.shape[-2]]
@@ -898,6 +899,13 @@ def forward(
         # See https://github.com/huggingface/transformers/pull/29402
         normalizer = torch.tensor(self.config.hidden_size**0.5, dtype=hidden_states.dtype)
         hidden_states = hidden_states * normalizer
+        if use_cache and not isinstance(past_key_values, Cache):  # kept for BC (non `Cache` `past_key_values` inputs)
+            return_legacy_cache = True
+            past_key_values = DynamicCache.from_legacy_cache(past_key_values)
+            logger.warning_once(
+                "We detected that you are passing `past_key_values` as a tuple and this is deprecated and will be removed in v4.43. "
+                "Please use an appropriate `Cache` class (https://huggingface.co/docs/transformers/v4.41.3/en/internal/generation_utils#transformers.Cache)"
+            )
 
         # decoder layers
         all_hidden_states = () if output_hidden_states else None
@@ -1397,7 +1405,7 @@ def set_input_embeddings(self, value):
     @add_start_docstrings_to_model_forward(GEMMA_INPUTS_DOCSTRING)
     def forward(
         self,
-        input_ids: torch.LongTensor = None,
+        input_ids: Optional[torch.LongTensor] = None,
         attention_mask: Optional[torch.Tensor] = None,
         position_ids: Optional[torch.LongTensor] = None,
         past_key_values: Optional[List[torch.FloatTensor]] = None,
@@ -1407,7 +1415,7 @@ def forward(
         output_attentions: Optional[bool] = None,
         output_hidden_states: Optional[bool] = None,
         return_dict: Optional[bool] = None,
-    ) -> Union[Tuple, SequenceClassifierOutputWithPast]:
+    ) -> Union[Tuple, TokenClassifierOutput]:
         r"""
         labels (`torch.LongTensor` of shape `(batch_size,)`, *optional*):
             Labels for computing the sequence classification/regression loss. Indices should be in `[0, ...,