chore: import upstream snapshot with attribution

conversion/xverse.py

@@ -0,0 +1,90 @@
+from __future__ import annotations
+
+import re
+
+from typing import Iterable, TYPE_CHECKING
+
+if TYPE_CHECKING:
+    from torch import Tensor
+
+from .base import ModelBase, TextModel, gguf
+
+
+@ModelBase.register("XverseForCausalLM")
+class XverseModel(TextModel):
+    model_arch = gguf.MODEL_ARCH.XVERSE
+
+    def set_vocab(self):
+        assert (self.dir_model / "tokenizer.json").is_file()
+        dir_model = self.dir_model
+        hparams = self.hparams
+
+        tokens: list[bytes] = []
+        toktypes: list[int] = []
+
+        from transformers import AutoTokenizer
+        tokenizer = AutoTokenizer.from_pretrained(dir_model)
+        vocab_size = hparams.get("vocab_size", len(tokenizer.vocab))  # ty: ignore[unresolved-attribute]
+        # Since we are checking the maximum index, we need to ensure it's strictly less than vocab_size,
+        # because vocab_size is the count of items, and indexes start at 0.
+        max_vocab_index = max(tokenizer.get_vocab().values())  # ty: ignore[unresolved-attribute]
+        if max_vocab_index >= vocab_size:
+            raise ValueError("Vocabulary size exceeds expected maximum size.")
+
+        reverse_vocab: dict[int, str] = {id_: encoded_tok for encoded_tok, id_ in tokenizer.vocab.items()}  # ty: ignore[unresolved-attribute]
+        added_vocab = tokenizer.get_added_vocab()  # ty: ignore[unresolved-attribute]
+
+        for token_id in range(vocab_size):
+            token_text = reverse_vocab[token_id].encode('utf-8')
+            # replace "\x00" to string with length > 0
+            if token_text == b"\x00":
+                toktype = gguf.TokenType.BYTE  # special
+                token_text = f"<{token_text}>".encode('utf-8')
+            elif re.fullmatch(br"<0x[0-9A-Fa-f]{2}>", token_text):
+                toktype = gguf.TokenType.BYTE  # special
+            elif reverse_vocab[token_id] in added_vocab:
+                if tokenizer.added_tokens_decoder[token_id].special:  # ty: ignore[unresolved-attribute]
+                    toktype = gguf.TokenType.CONTROL
+                else:
+                    toktype = gguf.TokenType.USER_DEFINED
+            else:
+                toktype = gguf.TokenType.NORMAL
+
+            tokens.append(token_text)
+            toktypes.append(toktype)
+
+        self.gguf_writer.add_tokenizer_model("llama")
+        self.gguf_writer.add_tokenizer_pre("default")
+        self.gguf_writer.add_token_list(tokens)
+        self.gguf_writer.add_token_types(toktypes)
+
+        special_vocab = gguf.SpecialVocab(dir_model, n_vocab=len(tokens))
+        special_vocab.add_to_gguf(self.gguf_writer)
+
+    def set_gguf_parameters(self):
+        super().set_gguf_parameters()
+
+        self.gguf_writer.add_tensor_data_layout("Meta AI original pth")
+        self.gguf_writer.add_rope_dimension_count(self.hparams["hidden_size"] // self.hparams["num_attention_heads"])
+
+    def modify_tensors(self, data_torch: Tensor, name: str, bid: int | None) -> Iterable[tuple[str, Tensor]]:
+        head_count = self.hparams["num_attention_heads"]
+        head_count_kv = self.hparams.get("num_key_value_heads", head_count)
+
+        # HF models permute some of the tensors, so we need to undo that
+        if name.endswith("q_proj.weight"):
+            data_torch = self._reverse_hf_permute(data_torch, head_count, head_count)
+        if name.endswith("k_proj.weight"):
+            data_torch = self._reverse_hf_permute(data_torch, head_count, head_count_kv)
+
+        yield from super().modify_tensors(data_torch, name, bid)
+
+    def _reverse_hf_permute(self, weights: Tensor, n_head: int, n_kv_head: int | None = None) -> Tensor:
+        if n_kv_head is not None and n_head != n_kv_head:
+            n_head //= n_kv_head
+
+        return (
+            weights.reshape(n_head, 2, weights.shape[0] // n_head // 2, *weights.shape[1:])
+            .swapaxes(1, 2)
+            .reshape(weights.shape)
+        )