From f2e69d28c01303ca9dc79907f89ef120a6ac4a92 Mon Sep 17 00:00:00 2001 From: sharpHL <132747147+sharpHL@users.noreply.github.com> Date: Sun, 28 Jan 2024 16:00:30 +0800 Subject: llama : add support for Orion-14B (#5118) * add support for Orion-14B(https://huggingface.co/OrionStarAI/Orion-14B-Chat) * flake8 support * Update llama.cpp Co-authored-by: Georgi Gerganov * Update llama.cpp Co-authored-by: Georgi Gerganov * Update llama.cpp Co-authored-by: Georgi Gerganov * Update llama.cpp Co-authored-by: Georgi Gerganov * Update llama.cpp Co-authored-by: slaren * Update llama.cpp * Update llama.cpp --------- Co-authored-by: lixiaopu Co-authored-by: Georgi Gerganov Co-authored-by: slaren --- gguf-py/gguf/constants.py | 24 +++++++++++++++++++++++- 1 file changed, 23 insertions(+), 1 deletion(-) (limited to 'gguf-py/gguf/constants.py') diff --git a/gguf-py/gguf/constants.py b/gguf-py/gguf/constants.py index 2d9c33c7..f5c933a4 100644 --- a/gguf-py/gguf/constants.py +++ b/gguf-py/gguf/constants.py @@ -101,6 +101,7 @@ class MODEL_ARCH(IntEnum): PHI2 = auto() PLAMO = auto() CODESHELL = auto() + ORION = auto() class MODEL_TENSOR(IntEnum): @@ -151,6 +152,7 @@ MODEL_ARCH_NAMES: dict[MODEL_ARCH, str] = { MODEL_ARCH.PHI2: "phi2", MODEL_ARCH.PLAMO: "plamo", MODEL_ARCH.CODESHELL: "codeshell", + MODEL_ARCH.ORION: "orion", } TENSOR_NAMES: dict[MODEL_TENSOR, str] = { @@ -427,7 +429,23 @@ MODEL_TENSORS: dict[MODEL_ARCH, list[MODEL_TENSOR]] = { MODEL_TENSOR.FFN_NORM, MODEL_TENSOR.FFN_DOWN, MODEL_TENSOR.FFN_UP, - ] + ], + MODEL_ARCH.ORION: [ + MODEL_TENSOR.TOKEN_EMBD, + MODEL_TENSOR.OUTPUT_NORM, + MODEL_TENSOR.OUTPUT, + MODEL_TENSOR.ROPE_FREQS, + MODEL_TENSOR.ATTN_NORM, + MODEL_TENSOR.ATTN_Q, + MODEL_TENSOR.ATTN_K, + MODEL_TENSOR.ATTN_V, + MODEL_TENSOR.ATTN_OUT, + MODEL_TENSOR.ATTN_ROT_EMBD, + MODEL_TENSOR.FFN_NORM, + MODEL_TENSOR.FFN_GATE, + MODEL_TENSOR.FFN_DOWN, + MODEL_TENSOR.FFN_UP, + ], # TODO } @@ -452,6 +470,10 @@ MODEL_TENSOR_SKIP: dict[MODEL_ARCH, list[MODEL_TENSOR]] = { MODEL_TENSOR.ROPE_FREQS, MODEL_TENSOR.ATTN_ROT_EMBD, ], + MODEL_ARCH.ORION: [ + MODEL_TENSOR.ROPE_FREQS, + MODEL_TENSOR.ATTN_ROT_EMBD, + ], } # -- cgit v1.2.3