NGRAMSPLIT(1)

✨ NAME ✨

ngramsplit — bestie, this is the manual page for ngramsplit 1.3.17, and it's giving "tiny tool, big job." 💅

---

🎀 DESCRIPTION 🎀

okay so here's the whole vibe: ngramsplit splits an n-gram model using context patterns. that's it. that's the mission. iconic behavior honestly. 🦋

here's how you call it:

Usage: ngramsplit [--options] in_fst [out_fsts_prefix]

now let's get into the flags, because there are a lot of them and we're going through every single one, no skips, no skimming, we're finishing the whole list like the main characters we are. 👑

🌸 PROGRAM FLAGS 🌸

• --backoff_label — type = int64_t, default = 0. this is the backoff label. that's the whole description, she's lowkey mysterious. 🎀
• --complete — type = bool, default = false. turn this on and it completes partial models.
• --contexts — type = std::string, default = "". the context patterns file.
• --far_type — type = std::string, default = "". the type of far to compile — and note the lil parenthetical, bestie: not FAR if empty. one of: "default", "stlist", "sttable".
• --method — type = std::string, default = "count_split". one of: "count_split", "histogram_split".
• --norm_eps — type = double, default = 0.001. the normalization check epsilon. ✨

---

💜 LIBRARY FLAGS 💜

🌟 Flags from: compile-strings.cc

• --far_field_separator — type = std::string, default = " + a literal newline + " (fr fr, the page prints an opening quote, a line break, and a closing quote — that default is a newline character, not a typo). the set of characters used as a separator between printed fields.

🌟 Flags from: flags.cc

• --help — type = bool, default = false. show usage information.
• --helpshort — type = bool, default = false. show brief usage information.
• --tmpdir — type = std::string, default = "/tmp". the temporary directory.
• --v — type = int32_t, default = 0. verbosity level.

🌟 Flags from: fst.cc

• --fst_align — type = bool, default = false. write FST data aligned where appropriate.
• --fst_default_cache_gc — type = bool, default = true. enable garbage collection of cache.
• --fst_default_cache_gc_limit — type = int64_t, default = 1048576. the cache byte size that triggers garbage collection.
• --fst_read_mode — type = std::string, default = "read". the default file reading mode for mappable files.
• --fst_verify_properties — type = bool, default = false. verify the FST properties queried by TestProperties.
• --save_relabel_ipairs — type = std::string, default = "". save input relabel pairs to file.
• --save_relabel_opairs — type = std::string, default = "". save output relabel pairs to file.

🌟 Flags from: ngram-output.cc

• --end_symbol — type = std::string, default = "</s>". the class label for sentence end.
• --start_symbol — type = std::string, default = "<s>". the class label for sentence start.

🌟 Flags from: symbol-table.cc

• --fst_compat_symbols — type = bool, default = true. require symbol tables to match when appropriate.
• --fst_field_separator — type = std::string, default = " " (yes bestie, that default is literally just a run of spaces, the page really does print it like that 🙈). the set of characters used as a separator between printed fields.

🌟 Flags from: util.cc

• --fst_error_fatal — type = bool, default = true. FST errors are fatal; otherwise (o.w.) return objects flagged as bad: e.g., FSTs: kError property set, FST weights: not a Member().
• --ngram_error_fatal — type = bool, default = true. NGram errors are fatal if true; otherwise it returns objects flagged as bad: e.g., NGramModel::Error() is true.

⚠️ heads up bestie: those two defaults are true, so by default errors are fatal. flip them off at your own risk, no cap.

🌟 Flags from: weight.cc

• --fst_weight_parentheses — type = std::string, default = "". the characters enclosing the first weight of a printed composite weight (e.g., pair weight, tuple weight and derived classes) to ensure proper I/O of nested composite weights; it must have size 0 (none) or 2 (open and close parenthesis).
• --fst_weight_separator — type = std::string, default = ",". the character separator between printed composite weights; it must be a single character.

---

that's the whole page, every last flag accounted for, we ate and left no crumbs. 🫶🌸

this has been ngramsplit 1.3.17, from February 2022 — stay hydrated, read your man pages, and slay the n-grams. 💖✨🎀