Skip to contents

LLMR 0.8.10

A bug-fix release; all changes are backward compatible.

  • llm_hash() no longer depends on the collation locale, so the same object hashes identically on every machine; hashes recorded under the C locale are unchanged.
  • Constant prompts (no {column} reference) now fill every row; previously rows 2..n were sent with NA content and rejected by providers.
  • llm_mutate() appends its generated columns again (instead of moving them to the front), and .before/.after now work on the embedding, structured, tag, and row-batched paths.
  • llm_usage() and llm_failures() work on llm_mutate_structured() and llm_mutate_tags() results; config-side and log-side request hashes agree after provider parameter renames.
  • Streaming no longer errors for providers outside the built-in table and now applies the req_builder hook and the request timeout. Assorted smaller fixes; see the commit log.

LLMR 0.8.9

LLMR 0.8.8

LLMR 0.8.7

LLMR 0.8.6

CRAN release: 2026-06-17

  • Row-packing controls renamed (.batch_size -> .rows_per_prompt, and kin); the word “batch” is now reserved for the asynchronous provider Batch API.
  • llm_log_read(): parse a JSONL audit log into records plus a per-record manifest with record and request hashes.
  • llm_request_hash() canonicalizes message shape and keys on all generation parameters, so a logged call and its config hash identically.
  • Added the shared reset() generic.

LLMR 0.8.5

LLMR 0.8.4

LLMR 0.8.3

LLMR 0.8.0

  • llm_preview(): render exactly what a call would send, and flag problems, without any API call.
  • llm_usage() and llm_failures(): outcome counts, token totals, and per-row failure listings for any result frame.
  • Missing token usage is reported as NA, not 0; a bare environment-variable name passed as api_key is always treated as a reference.
  • New “LLMR in 5 minutes” quickstart vignette, runnable on an open-weight model.

LLMR 0.7.2

  • Retry and error-classification fixes (provider errors containing braces, typed re-raise after exhausted retries).
  • llm_par_resume() re-runs only the failed rows; JSON array recovery and embedding-dimension fixes.

LLMR 0.7.1

  • Row batching for generative calls: pack several rows into one request, with fault-tolerant recovery for dropped, reordered, or truncated rows.

LLMR 0.7.0

LLMR 0.6.3

CRAN release: 2025-10-11

  • Ollama provider (local generative and embedding models); stable embedding column names.

LLMR 0.6.2

  • llm_mutate() shorthand (answer = "{question}") and .structured = TRUE.

LLMR 0.6.1

  • Fixed a bug affecting Anthropic calls.

LLMR 0.6.0

CRAN release: 2025-08-26

  • call_llm() returns an llmr_response object; as.character(x) extracts the text.
  • Secure API-key handling, structured JSON output, and multi-column injection in llm_mutate().