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 →
982 B25 linesraw
1from __future__ import annotations
2
3from .base import ModelBase, TextModel, gguf
4
5
6@ModelBase.register("GPTBigCodeForCausalLM")
7@ModelBase.example("bigcode/gpt_bigcode-santacoder")
8class StarCoderModel(TextModel):
9    model_arch = gguf.MODEL_ARCH.STARCODER
10
11    def set_gguf_parameters(self):
12        self.gguf_writer.add_context_length(self.hparams["n_positions"])
13        self.gguf_writer.add_embedding_length(self.hparams["n_embd"])
14        self.gguf_writer.add_feed_forward_length(4 * self.hparams["n_embd"])
15        self.gguf_writer.add_block_count(self.block_count)
16        self.gguf_writer.add_head_count(self.hparams["n_head"])
17        self.gguf_writer.add_head_count_kv(1)
18        self.gguf_writer.add_layer_norm_eps(self.hparams["layer_norm_epsilon"])
19        self.gguf_writer.add_file_type(self.ftype)
20
21
22@ModelBase.register("Starcoder2ForCausalLM")
23@ModelBase.example("bigcode/starcoder2-3b")
24class StarCoder2Model(TextModel):
25    model_arch = gguf.MODEL_ARCH.STARCODER2