summarize_traces_df() totals tokens, cost and latency over a traces data
frame. It is the data frame version of summarize_traces(), which takes a
module; use it for traces exported with export_traces() or converted from
vitals with as_dsprrr_traces().
Value
A dsprrr_trace_summary list with n_traces, total_tokens,
total_input_tokens, total_output_tokens, total_cost,
total_latency_ms, avg_latency_ms, avg_tokens_per_request,
token_breakdown (input, output and their ratio) and model_usage (a
data frame of requests per model).
See also
Other integrations:
as_dsprrr_metric(),
as_dsprrr_traces(),
as_ellmer_tool(),
as_vitals_cost(),
as_vitals_samples(),
as_vitals_solver(),
as_vitals_task(),
create_search_tool(),
llm_predict(),
ragnar_tool(),
reasoning_effort(),
register_dsprrr_engine(),
temperature(),
top_p(),
use_dsprrr_template(),
validate_workflow(),
vitals_metrics
Examples
traces <- data.frame(
model = c("gpt-6-luna", "gpt-6-luna"),
input_tokens = c(52L, 55L),
output_tokens = c(6L, 7L),
total_tokens = c(58L, 62L),
latency_ms = c(820, 640),
cost = c(0.0001, 0.0001)
)
summary <- summarize_traces_df(traces)
summary$total_tokens
#> [1] 120
summary$model_usage
#> model n_requests
#> 1 gpt-6-luna 2