Module model_files 

SafeTensors and GGUF model files as a table of their tensors. Only the header is read, so a 70 GB checkpoint opens like a 7 MB one. Both parsers are hand-written over a bounded reader, checking every stated length before allocating or skipping, so a hostile header errors rather than allocating gigabytes or panicking. Rows are a small eager frame made lazy; metadata and totals go to a ModelSummary for Info’s Model tab.

Structs

Header

One file’s header.

ModelSummary

What a model’s header says beyond its tensors, and their totals.

Remote

How a remote model’s files are reached: a source per URL and sibling file URLs. open and stop are called from concurrent shard readers (SHARD_READS).

Tensor

One tensor, as a header describes it.

TypeShare

One dtype’s share of a model.

Enums

MetaValue

One metadata value, as the Model tab shows it.

ModelKind

Which of the two formats a model file is.

RangeError

Why a ranged read stopped.

Constants

FIRST_GGUF_RANGE

The first GGUF header read; each next is twice the last up to MAX_RANGE, so a few KB costs one request and a 5-10 MB vocabulary four or five (see a_vocabulary_sized_gguf_header_takes_a_few_ranges).

FIRST_SAFETENSORS_RANGE

The first SafeTensors read: the header length and usually all its JSON (shard headers are KB to tens of KB); longer takes one more request.

MAX_SAFETENSORS_HEADER

The largest SafeTensors header read: the limit the reference implementation sets.

SHARD_READS

Shard headers read at once: hundreds of shards one at a time add up to minutes; a few at once gets most of the gain without a burst.

Traits

RangeSource

Bytes of one remote object, fetched a range at a time: an HTTP server, or an object in a store.

Functions

build

The table and the summary for headers already read; names are their files.

detail

The Model tab: the model’s totals, then its metadata as key and value, every value whole.

is_safetensors_index

Whether path is a SafeTensors index: model.safetensors.index.json.

looks_like_gguf

Whether the first bytes of a file are GGUF’s magic.

looks_like_safetensors

Whether the first bytes of a file are a SafeTensors header: a little-endian length a header could have, then the { that opens its JSON.

read_gguf

Read one GGUF header (versions 2 and 3, either byte order) from reader, which holds len bytes in all.

read_header_ranged

Read one model header from src with ranged requests: SafeTensors’ first FIRST_SAFETENSORS_RANGE plus any remaining JSON; GGUF in growing ranges until the tensor infos end. File-reader bounds all hold; stop is checked before each request.

read_header_ranged_from

As read_header_ranged with first range first (at least 8 bytes): small for the fuzz target and tests, to cross many ranges.

read_model

Read paths — model files, or SafeTensors indexes that name them — as one table of tensors, with a file column when there is more than one file.

read_remote_model

Read urls (remote model files, or SafeTensors indexes naming them) as one tensor table, fetching only headers, as read_model does on disk. RangeError::NoRanges only for one standalone file (downloadable instead).

read_safetensors

Read one SafeTensors header from reader, which holds len bytes in all.

url_file_name

The last segment of a URL, without a query: what a file’s row and its errors call it.

Type Aliases

Metadata

Metadata as key and value, in the order the file has them.

OpenRanges

A ranged source for a URL.