fix: "check out" as verb (#38678)

"check out" as verb

examples/modular-transformers/configuration_my_new_model.py

@@ -36,7 +36,7 @@ class MyNewModelConfig(PretrainedConfig):
             `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
+            by meanpooling all the original heads within that group. For more details, check out [this
             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"`):

examples/modular-transformers/configuration_new_model.py

@@ -34,7 +34,7 @@ class NewModelConfig(PretrainedConfig):
             `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
+            by meanpooling all the original heads within that group. For more details, check out [this
             paper](https://arxiv.org/pdf/2305.13245.pdf). If it is not specified, will default to
             `num_attention_heads`.
         head_dim (`int`, *optional*, defaults to 256):

src/transformers/models/aria/configuration_aria.py

@@ -49,7 +49,7 @@ class AriaTextConfig(PretrainedConfig):
             `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
+            by meanpooling all the original heads within that group. For more details, check out [this
             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"`):

src/transformers/models/aria/modular_aria.py

@@ -120,7 +120,7 @@ class AriaTextConfig(LlamaConfig):
             `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
+            by meanpooling all the original heads within that group. For more details, check out [this
             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"`):

src/transformers/models/bamba/configuration_bamba.py

@@ -53,7 +53,7 @@ class BambaConfig(PretrainedConfig):
             `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
+            by meanpooling all the original heads within that group. For more details, check out [this
             paper](https://arxiv.org/pdf/2305.13245.pdf). If it is not specified, will default to `8`.
         hidden_act (`str` or `function`, *optional*, defaults to `"silu"`):
             The non-linear activation function (function or string) in the decoder.

src/transformers/models/bitnet/configuration_bitnet.py

@@ -48,7 +48,7 @@ class BitNetConfig(PretrainedConfig):
             `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
+            by meanpooling all the original heads within that group. For more details, check out [this
             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 `"relu2"`):

src/transformers/models/chameleon/configuration_chameleon.py

@@ -125,7 +125,7 @@ class ChameleonConfig(PretrainedConfig):
             `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
+            by meanpooling all the original heads within that group. For more details, check out [this
             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"`):

src/transformers/models/cohere/configuration_cohere.py

@@ -56,7 +56,7 @@ class CohereConfig(PretrainedConfig):
             `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
+            by meanpooling all the original heads within that group. For more details, check out [this
             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"`):

src/transformers/models/cohere2/configuration_cohere2.py

@@ -52,7 +52,7 @@ class Cohere2Config(PretrainedConfig):
             `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
+            by meanpooling all the original heads within that group. For more details, check out [this
             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"`):

src/transformers/models/cohere2/modular_cohere2.py

@@ -74,7 +74,7 @@ class Cohere2Config(PretrainedConfig):
             `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
+            by meanpooling all the original heads within that group. For more details, check out [this
             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"`):

src/transformers/models/csm/configuration_csm.py

@@ -54,7 +54,7 @@ class CsmDepthDecoderConfig(PretrainedConfig):
             `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
+            by meanpooling all the original heads within that group. For more details, check out [this
             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"`):
@@ -235,7 +235,7 @@ class CsmConfig(PretrainedConfig):
             `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
+            by meanpooling all the original heads within that group. For more details, check out [this
             paper](https://arxiv.org/pdf/2305.13245.pdf).
         hidden_act (`str` or `function`, *optional*, defaults to `"silu"`):
             The non-linear activation function (function or string) in the backbone model Transformer decoder.

src/transformers/models/deepseek_v3/configuration_deepseek_v3.py

@@ -52,7 +52,7 @@ class DeepseekV3Config(PretrainedConfig):
             `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
+            by meanpooling all the original heads within that group. For more details, check out [this
             paper](https://arxiv.org/pdf/2305.13245.pdf). If it is not specified, will default to
             `num_attention_heads`.
         n_shared_experts (`int`, *optional*, defaults to 1):

src/transformers/models/diffllama/configuration_diffllama.py

@@ -48,7 +48,7 @@ class DiffLlamaConfig(PretrainedConfig):
             `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
+            by meanpooling all the original heads within that group. For more details, check out [this
             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"`):

src/transformers/models/emu3/configuration_emu3.py

@@ -138,7 +138,7 @@ class Emu3TextConfig(PretrainedConfig):
             `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
+            by meanpooling all the original heads within that group. For more details, check out [this
             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"`):

src/transformers/models/falcon_h1/configuration_falcon_h1.py

@@ -50,7 +50,7 @@ class FalconH1Config(PretrainedConfig):
             `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
+            by meanpooling all the original heads within that group. For more details, check out [this
             paper](https://arxiv.org/pdf/2305.13245.pdf). If it is not specified, will default to `8`.
         hidden_act (`str` or `function`, *optional*, defaults to `"silu"`):
             The non-linear activation function (function or string) in the decoder.

src/transformers/models/gemma/configuration_gemma.py

@@ -47,7 +47,7 @@ class GemmaConfig(PretrainedConfig):
             `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
+            by meanpooling all the original heads within that group. For more details, check out [this
             paper](https://arxiv.org/pdf/2305.13245.pdf). If it is not specified, will default to
             `num_attention_heads`.
         head_dim (`int`, *optional*, defaults to 256):

src/transformers/models/gemma/modular_gemma.py

@@ -74,7 +74,7 @@ class GemmaConfig(PretrainedConfig):
             `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
+            by meanpooling all the original heads within that group. For more details, check out [this
             paper](https://arxiv.org/pdf/2305.13245.pdf). If it is not specified, will default to
             `num_attention_heads`.
         head_dim (`int`, *optional*, defaults to 256):

src/transformers/models/gemma2/configuration_gemma2.py

@@ -47,7 +47,7 @@ class Gemma2Config(PretrainedConfig):
             `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
+            by meanpooling all the original heads within that group. For more details, check out [this
             paper](https://arxiv.org/pdf/2305.13245.pdf). If it is not specified, will default to
             `num_attention_heads`.
         head_dim (`int`, *optional*, defaults to 256):

src/transformers/models/gemma2/modular_gemma2.py

@@ -71,7 +71,7 @@ class Gemma2Config(PretrainedConfig):
             `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
+            by meanpooling all the original heads within that group. For more details, check out [this
             paper](https://arxiv.org/pdf/2305.13245.pdf). If it is not specified, will default to
             `num_attention_heads`.
         head_dim (`int`, *optional*, defaults to 256):

src/transformers/models/gemma3/configuration_gemma3.py

@@ -55,7 +55,7 @@ class Gemma3TextConfig(PretrainedConfig):
             `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
+            by meanpooling all the original heads within that group. For more details, check out [this
             paper](https://arxiv.org/pdf/2305.13245.pdf). If it is not specified, will default to
             `num_attention_heads`.
         head_dim (`int`, *optional*, defaults to 256):

src/transformers/models/gemma3/modular_gemma3.py

@@ -82,7 +82,7 @@ class Gemma3TextConfig(Gemma2Config, PretrainedConfig):
             `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
+            by meanpooling all the original heads within that group. For more details, check out [this
             paper](https://arxiv.org/pdf/2305.13245.pdf). If it is not specified, will default to
             `num_attention_heads`.
         head_dim (`int`, *optional*, defaults to 256):

src/transformers/models/glm/configuration_glm.py

@@ -42,7 +42,7 @@ class GlmConfig(PretrainedConfig):
             `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
+            by meanpooling all the original heads within that group. For more details, check out [this
             paper](https://arxiv.org/pdf/2305.13245.pdf). If it is not specified, will default to
             `num_attention_heads`.
         partial_rotary_factor (`float`, *optional*, defaults to 0.5): The factor of the partial rotary position.

src/transformers/models/glm4/configuration_glm4.py

@@ -42,7 +42,7 @@ class Glm4Config(PretrainedConfig):
             `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
+            by meanpooling all the original heads within that group. For more details, check out [this
             paper](https://arxiv.org/pdf/2305.13245.pdf). If it is not specified, will default to
             `num_attention_heads`.
         partial_rotary_factor (`float`, *optional*, defaults to 0.5): The factor of the partial rotary position.

src/transformers/models/granite/configuration_granite.py

@@ -54,7 +54,7 @@ class GraniteConfig(PretrainedConfig):
             `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
+            by meanpooling all the original heads within that group. For more details, check out [this
             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"`):

src/transformers/models/granitemoe/configuration_granitemoe.py

@@ -54,7 +54,7 @@ class GraniteMoeConfig(PretrainedConfig):
             `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
+            by meanpooling all the original heads within that group. For more details, check out [this
             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"`):

src/transformers/models/granitemoehybrid/configuration_granitemoehybrid.py

@@ -49,7 +49,7 @@ class GraniteMoeHybridConfig(PretrainedConfig):
             `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
+            by meanpooling all the original heads within that group. For more details, check out [this
             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"`):

src/transformers/models/granitemoeshared/configuration_granitemoeshared.py

@@ -54,7 +54,7 @@ class GraniteMoeSharedConfig(PretrainedConfig):
             `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
+            by meanpooling all the original heads within that group. For more details, check out [this
             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"`):

src/transformers/models/helium/configuration_helium.py

@@ -42,7 +42,7 @@ class HeliumConfig(PretrainedConfig):
             `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
+            by meanpooling all the original heads within that group. For more details, check out [this
             paper](https://arxiv.org/pdf/2305.13245.pdf). If it is not specified, will default to
             `num_attention_heads`.
         head_dim (`int`, *optional*, defaults to 128):

src/transformers/models/jamba/configuration_jamba.py

@@ -55,7 +55,7 @@ class JambaConfig(PretrainedConfig):
             `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
+            by meanpooling all the original heads within that group. For more details, check out [this
             paper](https://arxiv.org/pdf/2305.13245.pdf). If it is not specified, will default to `8`.
         hidden_act (`str` or `function`, *optional*, defaults to `"silu"`):
             The non-linear activation function (function or string) in the decoder.

src/transformers/models/llama/configuration_llama.py

@@ -51,7 +51,7 @@ class LlamaConfig(PretrainedConfig):
             `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
+            by meanpooling all the original heads within that group. For more details, check out [this
             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"`):

src/transformers/models/mimi/configuration_mimi.py

@@ -95,7 +95,7 @@ class MimiConfig(PretrainedConfig):
             `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
+            by meanpooling all the original heads within that group. For more details, check out [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.

src/transformers/models/minimax/configuration_minimax.py

@@ -51,7 +51,7 @@ class MiniMaxConfig(PretrainedConfig):
             `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
+            by meanpooling all the original heads within that group. For more details, check out [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.

src/transformers/models/minimax/modular_minimax.py

@@ -76,7 +76,7 @@ class MiniMaxConfig(MixtralConfig):
             `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
+            by meanpooling all the original heads within that group. For more details, check out [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.

src/transformers/models/mistral/configuration_mistral.py

@@ -51,7 +51,7 @@ class MistralConfig(PretrainedConfig):
             `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
+            by meanpooling all the original heads within that group. For more details, check out [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.

src/transformers/models/mixtral/configuration_mixtral.py

@@ -51,7 +51,7 @@ class MixtralConfig(PretrainedConfig):
             `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
+            by meanpooling all the original heads within that group. For more details, check out [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.

src/transformers/models/moonshine/configuration_moonshine.py

@@ -53,7 +53,7 @@ class MoonshineConfig(PretrainedConfig):
             `encoder_num_key_value_heads=encoder_num_attention_heads`, the model will use Multi Head Attention (MHA), if
             `encoder_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
+            by meanpooling all the original heads within that group. For more details, check out [this
             paper](https://arxiv.org/pdf/2305.13245.pdf). If it is not specified, will default to
             `num_attention_heads`.
         decoder_num_key_value_heads (`int`, *optional*):
@@ -61,7 +61,7 @@ class MoonshineConfig(PretrainedConfig):
             `decoder_num_key_value_heads=decoder_num_attention_heads`, the model will use Multi Head Attention (MHA), if
             `decoder_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
+            by meanpooling all the original heads within that group. For more details, check out [this
             paper](https://arxiv.org/pdf/2305.13245.pdf). If it is not specified, will default to
             `decoder_num_attention_heads`.
         pad_head_dim_to_multiple_of (`int`, *optional*):

src/transformers/models/moonshine/modular_moonshine.py

@@ -75,7 +75,7 @@ class MoonshineConfig(PretrainedConfig):
             `encoder_num_key_value_heads=encoder_num_attention_heads`, the model will use Multi Head Attention (MHA), if
             `encoder_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
+            by meanpooling all the original heads within that group. For more details, check out [this
             paper](https://arxiv.org/pdf/2305.13245.pdf). If it is not specified, will default to
             `num_attention_heads`.
         decoder_num_key_value_heads (`int`, *optional*):
@@ -83,7 +83,7 @@ class MoonshineConfig(PretrainedConfig):
             `decoder_num_key_value_heads=decoder_num_attention_heads`, the model will use Multi Head Attention (MHA), if
             `decoder_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
+            by meanpooling all the original heads within that group. For more details, check out [this
             paper](https://arxiv.org/pdf/2305.13245.pdf). If it is not specified, will default to
             `decoder_num_attention_heads`.
         pad_head_dim_to_multiple_of (`int`, *optional*):

src/transformers/models/moshi/configuration_moshi.py

@@ -47,7 +47,7 @@ class MoshiDepthConfig(PretrainedConfig):
             `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
+            by meanpooling all the original heads within that group. For more details, check out [this
             paper](https://arxiv.org/pdf/2305.13245.pdf). If it is not specified, will default to `num_attention_heads`.
         audio_vocab_size (`int`, *optional*, defaults to 2048):
             Vocabulary size of the audio part of model. Defines the number of different tokens that can be
@@ -171,7 +171,7 @@ class MoshiConfig(PretrainedConfig):
             `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
+            by meanpooling all the original heads within that group. For more details, check out [this
             paper](https://arxiv.org/pdf/2305.13245.pdf). If it is not specified, will default to `num_attention_heads`.
         audio_vocab_size (`int`, *optional*):
             Vocabulary size of the audio part of model. Defines the number of different tokens that can be

src/transformers/models/nemotron/configuration_nemotron.py

@@ -52,7 +52,7 @@ class NemotronConfig(PretrainedConfig):
             `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
+            by meanpooling all the original heads within that group. For more details, check out [this
             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 `"relu2"`):

src/transformers/models/olmo/configuration_olmo.py

@@ -53,7 +53,7 @@ class OlmoConfig(PretrainedConfig):
             `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
+            by meanpooling all the original heads within that group. For more details, check out [this
             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"`):

src/transformers/models/olmo2/configuration_olmo2.py

@@ -35,7 +35,7 @@ class Olmo2Config(PretrainedConfig):
             `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
+            by meanpooling all the original heads within that group. For more details, check out [this
             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"`):

src/transformers/models/olmo2/modular_olmo2.py

@@ -49,7 +49,7 @@ class Olmo2Config(OlmoConfig):
             `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
+            by meanpooling all the original heads within that group. For more details, check out [this
             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"`):

src/transformers/models/olmoe/configuration_olmoe.py

@@ -42,7 +42,7 @@ class OlmoeConfig(PretrainedConfig):
             `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
+            by meanpooling all the original heads within that group. For more details, check out [this
             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"`):

src/transformers/models/phi/configuration_phi.py

@@ -50,7 +50,7 @@ class PhiConfig(PretrainedConfig):
             `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
+            by meanpooling all the original heads within that group. For more details, check out [this
             paper](https://arxiv.org/pdf/2305.13245.pdf). If it is not specified, will default to
             `num_attention_heads`.
         resid_pdrop (`float`, *optional*, defaults to 0.0):

src/transformers/models/phi3/configuration_phi3.py

@@ -49,7 +49,7 @@ class Phi3Config(PretrainedConfig):
             `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
+            by meanpooling all the original heads within that group. For more details, check out [this
             paper](https://arxiv.org/pdf/2305.13245.pdf). If it is not specified, will default to
             `num_attention_heads`.
         resid_pdrop (`float`, *optional*, defaults to 0.0):

src/transformers/models/phi4_multimodal/configuration_phi4_multimodal.py

@@ -268,7 +268,7 @@ class Phi4MultimodalConfig(PretrainedConfig):
             `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
+            by meanpooling all the original heads within that group. For more details, check out [this
             paper](https://arxiv.org/pdf/2305.13245.pdf). If it is not specified, will default to
             `num_attention_heads`.
         resid_pdrop (`float`, *optional*, defaults to 0.0):

src/transformers/models/phi4_multimodal/modular_phi4_multimodal.py

@@ -304,7 +304,7 @@ class Phi4MultimodalConfig(Phi3Config):
             `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
+            by meanpooling all the original heads within that group. For more details, check out [this
             paper](https://arxiv.org/pdf/2305.13245.pdf). If it is not specified, will default to
             `num_attention_heads`.
         resid_pdrop (`float`, *optional*, defaults to 0.0):

src/transformers/models/phimoe/configuration_phimoe.py

@@ -48,7 +48,7 @@ class PhimoeConfig(PretrainedConfig):
             `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
+            by meanpooling all the original heads within that group. For more details, check out [this
             paper](https://arxiv.org/pdf/2305.13245.pdf). If it is not specified, will default to `8`.
         hidden_act (`str` or `function`, *optional*, defaults to `"silu"`):
             The non-linear activation function (function or string) in the decoder.

src/transformers/models/qwen2/configuration_qwen2.py

@@ -50,7 +50,7 @@ class Qwen2Config(PretrainedConfig):
             `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
+            by meanpooling all the original heads within that group. For more details, check out [this
             paper](https://arxiv.org/pdf/2305.13245.pdf). If it is not specified, will default to `32`.
         hidden_act (`str` or `function`, *optional*, defaults to `"silu"`):
             The non-linear activation function (function or string) in the decoder.

src/transformers/models/qwen2_5_omni/configuration_qwen2_5_omni.py

@@ -238,7 +238,7 @@ class Qwen2_5OmniTextConfig(PretrainedConfig):
             `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
+            by meanpooling all the original heads within that group. For more details, check out [this
             paper](https://arxiv.org/pdf/2305.13245.pdf). If it is not specified, will default to `32`.
         hidden_act (`str` or `function`, *optional*, defaults to `"silu"`):
             The non-linear activation function (function or string) in the decoder.
@@ -584,7 +584,7 @@ class Qwen2_5OmniTalkerConfig(PretrainedConfig):
             `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
+            by meanpooling all the original heads within that group. For more details, check out [this
             paper](https://arxiv.org/pdf/2305.13245.pdf). If it is not specified, will default to `32`.
         hidden_act (`str` or `function`, *optional*, defaults to `"silu"`):
             The non-linear activation function (function or string) in the decoder.

src/transformers/models/qwen2_5_omni/modular_qwen2_5_omni.py

@@ -277,7 +277,7 @@ class Qwen2_5OmniTextConfig(PretrainedConfig):
             `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
+            by meanpooling all the original heads within that group. For more details, check out [this
             paper](https://arxiv.org/pdf/2305.13245.pdf). If it is not specified, will default to `32`.
         hidden_act (`str` or `function`, *optional*, defaults to `"silu"`):
             The non-linear activation function (function or string) in the decoder.
@@ -623,7 +623,7 @@ class Qwen2_5OmniTalkerConfig(PretrainedConfig):
             `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
+            by meanpooling all the original heads within that group. For more details, check out [this
             paper](https://arxiv.org/pdf/2305.13245.pdf). If it is not specified, will default to `32`.
         hidden_act (`str` or `function`, *optional*, defaults to `"silu"`):
             The non-linear activation function (function or string) in the decoder.

src/transformers/models/qwen2_5_vl/configuration_qwen2_5_vl.py

@@ -94,7 +94,7 @@ class Qwen2_5_VLTextConfig(PretrainedConfig):
             `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
+            by meanpooling all the original heads within that group. For more details, check out [this
             paper](https://arxiv.org/pdf/2305.13245.pdf). If it is not specified, will default to `32`.
         hidden_act (`str` or `function`, *optional*, defaults to `"silu"`):
             The non-linear activation function (function or string) in the decoder.

src/transformers/models/qwen2_moe/configuration_qwen2_moe.py

@@ -49,7 +49,7 @@ class Qwen2MoeConfig(PretrainedConfig):
             `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
+            by meanpooling all the original heads within that group. For more details, check out [this
             paper](https://arxiv.org/pdf/2305.13245.pdf). If it is not specified, will default to `32`.
         hidden_act (`str` or `function`, *optional*, defaults to `"silu"`):
             The non-linear activation function (function or string) in the decoder.

src/transformers/models/qwen2_vl/configuration_qwen2_vl.py

@@ -83,7 +83,7 @@ class Qwen2VLTextConfig(PretrainedConfig):
             `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
+            by meanpooling all the original heads within that group. For more details, check out [this
             paper](https://arxiv.org/pdf/2305.13245.pdf). If it is not specified, will default to `32`.
         hidden_act (`str` or `function`, *optional*, defaults to `"silu"`):
             The non-linear activation function (function or string) in the decoder.

src/transformers/models/qwen3/configuration_qwen3.py

@@ -50,7 +50,7 @@ class Qwen3Config(PretrainedConfig):
             `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
+            by meanpooling all the original heads within that group. For more details, check out [this
             paper](https://arxiv.org/pdf/2305.13245.pdf). If it is not specified, will default to `32`.
         head_dim (`int`, *optional*, defaults to 128):
             The attention head dimension.

src/transformers/models/qwen3_moe/configuration_qwen3_moe.py

@@ -49,7 +49,7 @@ class Qwen3MoeConfig(PretrainedConfig):
             `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
+            by meanpooling all the original heads within that group. For more details, check out [this
             paper](https://arxiv.org/pdf/2305.13245.pdf). If it is not specified, will default to `32`.
         hidden_act (`str` or `function`, *optional*, defaults to `"silu"`):
             The non-linear activation function (function or string) in the decoder.

src/transformers/models/stablelm/configuration_stablelm.py

@@ -51,7 +51,7 @@ class StableLmConfig(PretrainedConfig):
             `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
+            by meanpooling all the original heads within that group. For more details, check out [this
             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"`):

src/transformers/models/starcoder2/configuration_starcoder2.py

@@ -50,7 +50,7 @@ class Starcoder2Config(PretrainedConfig):
             `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
+            by meanpooling all the original heads within that group. For more details, check out [this
             paper](https://arxiv.org/pdf/2305.13245.pdf). If it is not specified, will default to `8`.
         hidden_act (`str` or `function`, *optional*, defaults to `"gelu_pytorch_tanh"`):
             The non-linear activation function (function or string) in the decoder.

src/transformers/models/zamba/configuration_zamba.py

@@ -59,7 +59,7 @@ class ZambaConfig(PretrainedConfig):
             `num_key_value_heads=None`, 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
+            by meanpooling all the original heads within that group. For more details, check out [this
             paper](https://arxiv.org/pdf/2305.13245.pdf).
         n_mamba_heads (`int`, *optional*, defaults to 2):
             Number of mamba heads for each mamba layer.

src/transformers/models/zamba2/configuration_zamba2.py

@@ -79,7 +79,7 @@ class Zamba2Config(PretrainedConfig):
             `num_key_value_heads=None`, 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
+            by meanpooling all the original heads within that group. For more details, check out [this
             paper](https://arxiv.org/pdf/2305.13245.pdf).
         attention_dropout (`float`, *optional*, defaults to 0.0):
             The dropout ratio for the attention probabilities.

src/transformers/utils/hub.py

@@ -542,7 +542,7 @@ def cached_files(
             elif _raise_exceptions_for_missing_entries:
                 raise OSError(
                     f"We couldn't connect to '{HUGGINGFACE_CO_RESOLVE_ENDPOINT}' to load the files, and couldn't find them in the"
-                    f" cached files.\nCheckout your internet connection or see how to run the library in offline mode at"
+                    f" cached files.\nCheck your internet connection or see how to run the library in offline mode at"
                     " 'https://huggingface.co/docs/transformers/installation#offline-mode'."
                 ) from e
         # snapshot_download will not raise EntryNotFoundError, but hf_hub_download can. If this is the case, it will be treated

templates/adding_a_new_model/README.md

@@ -19,5 +19,5 @@ limitations under the License.
 This page has been updated in light of the removal of the `add_new_model` script in favor of the more complete 
 `add_new_model_like` script.
 
-We recommend you checkout the documentation of [How to add a model](https://huggingface.co/docs/transformers/main/en/add_new_model)
-in the Hugging Face Transformers documentation for complete and up-to-date instructions.
+We recommend you check out the documentation on [how to add a model](https://huggingface.co/docs/transformers/main/en/add_new_model) 
+for complete and up-to-date instructions.