Fix documentation typos (#31476)

Fix doc typo

src/transformers/models/cohere/configuration_cohere.py

@@ -53,7 +53,7 @@ class CohereConfig(PretrainedConfig):
         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
+            `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

src/transformers/models/gemma/configuration_gemma.py

@@ -47,7 +47,7 @@ class GemmaConfig(PretrainedConfig):
         num_key_value_heads (`int`, *optional*, defaults to 16):
             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
+            `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

src/transformers/models/gemma/diff_gemma.py

@@ -64,7 +64,7 @@ class GemmaConfig(PretrainedConfig):
         num_key_value_heads (`int`, *optional*, defaults to 16):
             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
+            `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

src/transformers/models/jamba/configuration_jamba.py

@@ -53,7 +53,7 @@ class JambaConfig(PretrainedConfig):
         num_key_value_heads (`int`, *optional*, defaults to 8):
             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
+            `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`.

src/transformers/models/llama/configuration_llama.py

@@ -51,7 +51,7 @@ class LlamaConfig(PretrainedConfig):
         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
+            `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

src/transformers/models/mistral/configuration_mistral.py

@@ -49,7 +49,7 @@ class MistralConfig(PretrainedConfig):
         num_key_value_heads (`int`, *optional*, defaults to 8):
             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
+            `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`.

src/transformers/models/mixtral/configuration_mixtral.py

@@ -49,7 +49,7 @@ class MixtralConfig(PretrainedConfig):
         num_key_value_heads (`int`, *optional*, defaults to 8):
             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
+            `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`.

src/transformers/models/olmo/configuration_olmo.py

@@ -51,7 +51,7 @@ class OlmoConfig(PretrainedConfig):
         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
+            `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

src/transformers/models/phi/configuration_phi.py

@@ -47,7 +47,7 @@ class PhiConfig(PretrainedConfig):
         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
+            `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

src/transformers/models/phi3/configuration_phi3.py

@@ -47,7 +47,7 @@ class Phi3Config(PretrainedConfig):
         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
+            `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

src/transformers/models/qwen2/configuration_qwen2.py

@@ -47,7 +47,7 @@ class Qwen2Config(PretrainedConfig):
         num_key_value_heads (`int`, *optional*, defaults to 32):
             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
+            `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 `32`.

src/transformers/models/qwen2_moe/configuration_qwen2_moe.py

@@ -47,7 +47,7 @@ class Qwen2MoeConfig(PretrainedConfig):
         num_key_value_heads (`int`, *optional*, defaults to 16):
             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
+            `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 `32`.

src/transformers/models/stablelm/configuration_stablelm.py

@@ -48,7 +48,7 @@ class StableLmConfig(PretrainedConfig):
         num_key_value_heads (`int`, *optional*, defaults to 32):
             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
+            `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

src/transformers/models/starcoder2/configuration_starcoder2.py

@@ -47,7 +47,7 @@ class Starcoder2Config(PretrainedConfig):
         num_key_value_heads (`int`, *optional*, defaults to 2):
             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
+            `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`.