[GGUF] Refactor and decouple gguf checkpoint loading logic (#34385)
* draft load_gguf refactor
* update
Signed-off-by: Isotr0py <2037008807@qq.com>
* remove llama mapping
Signed-off-by: Isotr0py <2037008807@qq.com>
* remove qwen2 mapping
Signed-off-by: Isotr0py <2037008807@qq.com>
* remove unused function
Signed-off-by: Isotr0py <2037008807@qq.com>
* deprecate stablelm mapping
Signed-off-by: Isotr0py <2037008807@qq.com>
* deprecate phi3 mapping
Signed-off-by: Isotr0py <2037008807@qq.com>
* deprecate t5 mapping
Signed-off-by: Isotr0py <2037008807@qq.com>
* deprecate bloom mapping
Signed-off-by: Isotr0py <2037008807@qq.com>
* fix bloom
Signed-off-by: Isotr0py <2037008807@qq.com>
* deprecate starcoder2 mapping
Signed-off-by: Isotr0py <2037008807@qq.com>
* deprecate gpt2 mapping
Signed-off-by: Isotr0py <2037008807@qq.com>
* deprecate mistral mapping
Signed-off-by: Isotr0py <2037008807@qq.com>
* deprecate nemotron mapping
Signed-off-by: Isotr0py <2037008807@qq.com>
* deprecate mamba mapping
Signed-off-by: Isotr0py <2037008807@qq.com>
* deprecate mamba mapping
Signed-off-by: Isotr0py <2037008807@qq.com>
* code format
Signed-off-by: Isotr0py <2037008807@qq.com>
* code format
Signed-off-by: Isotr0py <2037008807@qq.com>
* fix mamba
Signed-off-by: Isotr0py <2037008807@qq.com>
* fix qwen2moe
Signed-off-by: Isotr0py <2037008807@qq.com>
* remove qwen2moe mapping
Signed-off-by: Isotr0py <2037008807@qq.com>
* clean up
Signed-off-by: Isotr0py <2037008807@qq.com>
* remove falcon 7b map
Signed-off-by: Isotr0py <2037008807@qq.com>
* remove all ggml tensors mapping
Signed-off-by: Isotr0py <2037008807@qq.com>
* add comments
Signed-off-by: Isotr0py <2037008807@qq.com>
* update messages
Signed-off-by: Isotr0py <2037008807@qq.com>
* fix tensors in parsed parameters
Signed-off-by: Isotr0py <2037008807@qq.com>
* add gguf check
Signed-off-by: Isotr0py <2037008807@qq.com>
---------
Signed-off-by: Isotr0py <2037008807@qq.com>