Write an OpenAI batch table to a JSONL file

This helper takes the output of build_openai_batch_requests (or a compatible table) and writes one JSON object per line, in the format expected by the OpenAI batch API.

Usage

write_openai_batch_file(batch_tbl, path)

Arguments

batch_tbl

A data frame or tibble, typically the result of build_openai_batch_requests.

path

File path where the JSONL file should be written.

Value

Invisibly returns path.

Details

The input can either:

• Already contain a character column jsonl (one JSON string per row), in which case that column is used directly, or

• Contain the columns custom_id, method, url, and body, in which case the JSON strings are constructed automatically.

See also

llm_submit_pairs_batch(), llm_download_batch_results()

Other batch backends: anthropic_create_batch(), anthropic_download_batch_results(), anthropic_get_batch(), anthropic_poll_batch_until_complete(), build_anthropic_batch_requests(), build_gemini_batch_requests(), build_openai_batch_requests(), gemini_create_batch(), gemini_download_batch_results(), gemini_get_batch(), gemini_poll_batch_until_complete(), llm_download_batch_results(), llm_resume_multi_batches(), llm_submit_pairs_batch(), llm_submit_pairs_multi_batch(), openai_create_batch(), openai_download_batch_errors(), openai_download_batch_output(), openai_get_batch(), openai_poll_batch_until_complete(), openai_upload_batch_file(), run_anthropic_batch_pipeline(), run_gemini_batch_pipeline(), run_openai_batch_pipeline()

Examples

# Construct a minimal batch request tibble
requests <- tibble::tibble(
  custom_id = c("req1", "req2"),
  method = "POST",
  url = "/v1/chat/completions",
  body = list(
    list(
      model = "gpt-4o-mini",
      messages = list(
        list(role = "user", content = "Hello")
      )
    ),
    list(
      model = "gpt-4o-mini",
      messages = list(
        list(role = "user", content = "Goodbye")
      )
    )
  )
)

# Write to a temporary JSONL file
path <- tempfile(fileext = ".jsonl")
write_openai_batch_file(requests, path)

# Inspect the file contents
readLines(path)
#> [1] "{\"custom_id\":\"req1\",\"method\":\"POST\",\"url\":\"/v1/chat/completions\",\"body\":{\"model\":\"gpt-4o-mini\",\"messages\":[{\"role\":\"user\",\"content\":\"Hello\"}]}}"  
#> [2] "{\"custom_id\":\"req2\",\"method\":\"POST\",\"url\":\"/v1/chat/completions\",\"body\":{\"model\":\"gpt-4o-mini\",\"messages\":[{\"role\":\"user\",\"content\":\"Goodbye\"}]}}"