char-fork/llama.cpp

git clone https://git.t4t.associates/char-fork/llama.cpp

Xuan-Son Nguyenconvert: add @ModelBase.example (#27208)fa88ae936

char/aliceai-t5-moe
brancheschar/aliceai-t5-moeall refs →
835 B24 linesraw
1from __future__ import annotations
2
3from .base import ModelBase, TextModel, gguf
4
5
6@ModelBase.register("PLMForCausalLM")
7@ModelBase.example("PLM-Team/PLM-1.8B-Instruct")
8class PLMModel(TextModel):
9    model_arch = gguf.MODEL_ARCH.PLM
10
11    def set_vocab(self):
12        self._set_vocab_gpt2()
13
14    def set_gguf_parameters(self):
15        super().set_gguf_parameters()
16        hparams = self.hparams
17        self.gguf_writer.add_vocab_size(hparams["vocab_size"])
18        self.gguf_writer.add_kv_lora_rank(hparams["kv_lora_rank"])
19        self.gguf_writer.add_key_length(hparams["qk_nope_head_dim"] + hparams["qk_rope_head_dim"])
20        self.gguf_writer.add_value_length(hparams["v_head_dim"])
21        self.gguf_writer.add_rope_dimension_count(hparams["qk_rope_head_dim"])
22
23    def prepare_tensors(self):
24        super().prepare_tensors()