From 316c7faf7740fa98ea68f1445f4505810f706b9e Mon Sep 17 00:00:00 2001 From: runfuture Date: Wed, 7 Feb 2024 14:15:56 +0800 Subject: llama : add MiniCPM support (#5346) * support minicpm arch. * fix tab/space typo. * convert minicpm model via convert-hf-gguf.py * try to make tokenizer work * fix bug for quantize minicpm * fix for flake8 lint * remove convert-minicpm.py * fix for editorconfig * correct minicpm model type (size) * constants expanded for minicpm * Minor change of the constant names for minicpm --- gguf-py/gguf/constants.py | 21 +++++++++++++++++++++ 1 file changed, 21 insertions(+) (limited to 'gguf-py') diff --git a/gguf-py/gguf/constants.py b/gguf-py/gguf/constants.py index ed8e26f8..1cfd41c0 100644 --- a/gguf-py/gguf/constants.py +++ b/gguf-py/gguf/constants.py @@ -104,6 +104,7 @@ class MODEL_ARCH(IntEnum): CODESHELL = auto() ORION = auto() INTERNLM2 = auto() + MINICPM = auto() class MODEL_TENSOR(IntEnum): @@ -156,6 +157,7 @@ MODEL_ARCH_NAMES: dict[MODEL_ARCH, str] = { MODEL_ARCH.CODESHELL: "codeshell", MODEL_ARCH.ORION: "orion", MODEL_ARCH.INTERNLM2: "internlm2", + MODEL_ARCH.MINICPM: "minicpm", } TENSOR_NAMES: dict[MODEL_TENSOR, str] = { @@ -464,6 +466,25 @@ MODEL_TENSORS: dict[MODEL_ARCH, list[MODEL_TENSOR]] = { MODEL_TENSOR.FFN_DOWN, MODEL_TENSOR.FFN_UP, ], + MODEL_ARCH.MINICPM: [ + MODEL_TENSOR.TOKEN_EMBD, + MODEL_TENSOR.OUTPUT_NORM, + 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_GATE_INP, + MODEL_TENSOR.FFN_NORM, + MODEL_TENSOR.FFN_GATE, + MODEL_TENSOR.FFN_DOWN, + MODEL_TENSOR.FFN_UP, + MODEL_TENSOR.FFN_GATE_EXP, + MODEL_TENSOR.FFN_DOWN_EXP, + MODEL_TENSOR.FFN_UP_EXP, + ], # TODO } -- cgit v1.2.3