Update cognitivess_model/configuration_cognitivess.py

cognitivess_model/configuration_cognitivess.py

@@ -1,6 +1,5 @@
 # coding=utf-8
-# Copyright 2023 Cognitivess AI and the HuggingFace Inc. team. All rights reserved.
-#
+# Copyright 2022 Cognitivess and the HuggingFace Inc. team. All rights reserved.
 # Licensed under the Apache License, Version 2.0 (the "License");
 # you may not use this file except in compliance with the License.
 # You may obtain a copy of the License at
@@ -23,11 +22,9 @@ logger = logging.get_logger(__name__)
 
 class CognitivessConfig(PretrainedConfig):
     r"""
-    This is the configuration class to store the configuration of a [`CognitivessModel`]. It is used to instantiate an
-    Cognitivess model according to the specified arguments, defining the model architecture. Instantiating a configuration
-    with the defaults will yield a similar configuration to that of the Cognitivess-8B-v0.1.
-
-    [CognitivessAI/cognitivess](https://huggingface.co/CognitivessAI/cognitivess)
+    This is the configuration class to store the configuration of a [`CognitivessModel`]. It is used to instantiate an Cognitivess
+    model according to the specified arguments, defining the model architecture. Instantiating a configuration with the
+    defaults will yield a similar configuration to that of the Cognitivess.
 
     Configuration objects inherit from [`PretrainedConfig`] and can be used to control the model outputs. Read the
     documentation from [`PretrainedConfig`] for more information.
@@ -39,26 +36,25 @@ class CognitivessConfig(PretrainedConfig):
             `inputs_ids` passed when calling [`CognitivessModel`]
         hidden_size (`int`, *optional*, defaults to 4096):
             Dimension of the hidden representations.
-        intermediate_size (`int`, *optional*, defaults to 14336):
+        intermediate_size (`int`, *optional*, defaults to 11008):
             Dimension of the MLP representations.
         num_hidden_layers (`int`, *optional*, defaults to 32):
-            Number of hidden layers in the Transformer encoder.
+            Number of hidden layers in the Transformer decoder.
         num_attention_heads (`int`, *optional*, defaults to 32):
-            Number of attention heads for each attention layer in the Transformer encoder.
-        num_key_value_heads (`int`, *optional*, defaults to 8):
+            Number of attention heads for each attention layer in the Transformer decoder.
+        num_key_value_heads (`int`, *optional*):
             This is the number of key_value heads that should be used to implement Grouped Query Attention. If
             `num_key_value_heads=num_attention_heads`, the model will use Multi Head Attention (MHA), if
             `num_key_value_heads=1` the model will use Multi Query Attention (MQA) otherwise GQA is used. When
             converting a multi-head checkpoint to a GQA checkpoint, each group key and value head should be constructed
             by meanpooling all the original heads within that group. For more details checkout [this
-            paper](https://arxiv.org/pdf/2305.13245.pdf). If it is not specified, will default to `8`.
-        head_dim (`int`, *optional*, defaults to `hidden_size // num_attention_heads`):
-            The attention head dimension.
+            paper](https://arxiv.org/pdf/2305.13245.pdf). If it is not specified, will default to
+            `num_attention_heads`.
         hidden_act (`str` or `function`, *optional*, defaults to `"silu"`):
             The non-linear activation function (function or string) in the decoder.
-        max_position_embeddings (`int`, *optional*, defaults to `4096*32`):
-            The maximum sequence length that this model might ever be used with. Cognitivess's sliding window attention
-            allows sequence of up to 4096*32 tokens.
+        max_position_embeddings (`int`, *optional*, defaults to 2048):
+            The maximum sequence length that this model might ever be used with. Cognitivess 1 supports up to 2048 tokens,
+            Cognitivess 2 up to 4096, CodeCognitivess up to 16384.
         initializer_range (`float`, *optional*, defaults to 0.02):
             The standard deviation of the truncated_normal_initializer for initializing all weight matrices.
         rms_norm_eps (`float`, *optional*, defaults to 1e-06):
@@ -67,27 +63,42 @@ class CognitivessConfig(PretrainedConfig):
             Whether or not the model should return the last key/values attentions (not used by all models). Only
             relevant if `config.is_decoder=True`.
         pad_token_id (`int`, *optional*):
-            The id of the padding token.
+            Padding token id.
         bos_token_id (`int`, *optional*, defaults to 1):
-            The id of the "beginning-of-sequence" token.
+            Beginning of stream token id.
         eos_token_id (`int`, *optional*, defaults to 2):
-            The id of the "end-of-sequence" token.
+            End of stream token id.
+        pretraining_tp (`int`, *optional*, defaults to 1):
+            Experimental feature. Tensor parallelism rank used during pretraining. Please refer to [this
+            document](https://huggingface.co/docs/transformers/main/perf_train_gpu_many#tensor-parallelism) to understand more about it. This value is
+            necessary to ensure exact reproducibility of the pretraining results. Please refer to [this
+            issue](https://github.com/pytorch/pytorch/issues/76232).
         tie_word_embeddings (`bool`, *optional*, defaults to `False`):
-            Whether the model's input and output word embeddings should be tied.
+            Whether to tie weight embeddings
         rope_theta (`float`, *optional*, defaults to 10000.0):
             The base period of the RoPE embeddings.
-        sliding_window (`int`, *optional*, defaults to 4096):
-            Sliding window attention window size. If not specified, will default to `4096`.
+        rope_scaling (`Dict`, *optional*):
+            Dictionary containing the scaling configuration for the RoPE embeddings. Currently supports two scaling
+            strategies: linear and dynamic. Their scaling factor must be a float greater than 1. The expected format is
+            `{"type": strategy name, "factor": scaling factor}`. When using this flag, don't update
+            `max_position_embeddings` to the expected new maximum. See the following thread for more information on how
+            these scaling strategies behave:
+            https://www.reddit.com/r/LocalCognitivess/comments/14mrgpr/dynamically_scaled_rope_further_increases/. This is an
+            experimental feature, subject to breaking API changes in future versions.
+        attention_bias (`bool`, *optional*, defaults to `False`):
+            Whether to use a bias in the query, key, value and output projection layers during self-attention.
         attention_dropout (`float`, *optional*, defaults to 0.0):
             The dropout ratio for the attention probabilities.
+        mlp_bias (`bool`, *optional*, defaults to `False`):
+            Whether to use a bias in up_proj, down_proj and gate_proj layers in the MLP layers.
 
     ```python
     >>> from transformers import CognitivessModel, CognitivessConfig
 
-    >>> # Initializing a Cognitivess 8B style configuration
+    >>> # Initializing a Cognitivess Cognitivess style configuration
     >>> configuration = CognitivessConfig()
 
-    >>> # Initializing a model from the Cognitivess 8B style configuration
+    >>> # Initializing a model from the Cognitivess style configuration
     >>> model = CognitivessModel(configuration)
 
     >>> # Accessing the model configuration
@@ -101,23 +112,25 @@ class CognitivessConfig(PretrainedConfig):
         self,
         vocab_size=32000,
         hidden_size=4096,
-        intermediate_size=14336,
+        intermediate_size=11008,
         num_hidden_layers=32,
         num_attention_heads=32,
-        num_key_value_heads=8,
-        head_dim=None,
+        num_key_value_heads=None,
         hidden_act="silu",
-        max_position_embeddings=4096 * 32,
+        max_position_embeddings=2048,
         initializer_range=0.02,
         rms_norm_eps=1e-6,
         use_cache=True,
         pad_token_id=None,
         bos_token_id=1,
         eos_token_id=2,
+        pretraining_tp=1,
         tie_word_embeddings=False,
         rope_theta=10000.0,
-        sliding_window=4096,
+        rope_scaling=None,
+        attention_bias=False,
         attention_dropout=0.0,
+        mlp_bias=False,
         **kwargs,
     ):
         self.vocab_size = vocab_size
@@ -126,8 +139,6 @@ class CognitivessConfig(PretrainedConfig):
         self.intermediate_size = intermediate_size
         self.num_hidden_layers = num_hidden_layers
         self.num_attention_heads = num_attention_heads
-        self.sliding_window = sliding_window
-        self.head_dim = head_dim or hidden_size // num_attention_heads
 
         # for backward compatibility
         if num_key_value_heads is None:
@@ -137,9 +148,14 @@ class CognitivessConfig(PretrainedConfig):
         self.hidden_act = hidden_act
         self.initializer_range = initializer_range
         self.rms_norm_eps = rms_norm_eps
+        self.pretraining_tp = pretraining_tp
         self.use_cache = use_cache
         self.rope_theta = rope_theta
+        self.rope_scaling = rope_scaling
+        self._rope_scaling_validation()
+        self.attention_bias = attention_bias
         self.attention_dropout = attention_dropout
+        self.mlp_bias = mlp_bias
 
         super().__init__(
             pad_token_id=pad_token_id,
@@ -147,4 +163,24 @@ class CognitivessConfig(PretrainedConfig):
             eos_token_id=eos_token_id,
             tie_word_embeddings=tie_word_embeddings,
             **kwargs,
-        )
+        )
+
+    def _rope_scaling_validation(self):
+        """
+        Validate the `rope_scaling` configuration.
+        """
+        if self.rope_scaling is None:
+            return
+
+        if not isinstance(self.rope_scaling, dict) or len(self.rope_scaling) != 2:
+            raise ValueError(
+                "`rope_scaling` must be a dictionary with two fields, `type` and `factor`, " f"got {self.rope_scaling}"
+            )
+        rope_scaling_type = self.rope_scaling.get("type", None)
+        rope_scaling_factor = self.rope_scaling.get("factor", None)
+        if rope_scaling_type is None or rope_scaling_type not in ["linear", "dynamic"]:
+            raise ValueError(
+                f"`rope_scaling`'s type field must be one of ['linear', 'dynamic'], got {rope_scaling_type}"
+            )
+        if rope_scaling_factor is None or not isinstance(rope_scaling_factor, float) or rope_scaling_factor <= 1.0:
+            raise ValueError(f"`rope_scaling`'s factor field must be a float > 1, got {rope_scaling_factor}")