Documentation
¶
Index ¶
- Variables
- func ExportToFile(dataset []map[string]any, outputPath string)
- func ExportToLangSmith(ctx context.Context, sdk *langsmith.Client, dataset []map[string]any, ...)
- func ExtractDocuments(outputs map[string]any) []string
- func ExtractFinalOutput(runs []RunData, outputFields []string) string
- func ExtractFromMessages(messages []any, role string) string
- func ExtractToolSequence(runs []RunData, depth *int) []string
- func ExtractTraceInputs(root RunData, inputFields []string, asDict bool) any
- func ExtractTraceOutput(root RunData, outputFields []string, messagesOnly bool) string
- func ExtractValue(data map[string]any, fields []string, commonFields []string, ...) any
- func FindRetrievalData(runs []RunData) map[string]any
- func GenerateDataset(traces []Trace, datasetType string, runName string, depth *int, ...) []map[string]any
- func GetNodeIO(runs []RunData, runName string) []map[string]any
- type RunData
- type Trace
Constants ¶
This section is empty.
Variables ¶
var ( CommonInputFields = []string{"query", "input", "question", "message", "prompt", "text"} CommonOutputFields = []string{"answer", "output", "response", "result"} )
Common field names for heuristic extraction.
Functions ¶
func ExportToFile ¶
ExportToFile exports a generated dataset to a file (JSON or CSV).
func ExportToLangSmith ¶
func ExportToLangSmith(ctx context.Context, sdk *langsmith.Client, dataset []map[string]any, datasetName string, datasetType string)
ExportToLangSmith uploads a generated dataset to LangSmith.
func ExtractDocuments ¶
ExtractDocuments extracts document text from retriever outputs.
func ExtractFinalOutput ¶
ExtractFinalOutput searches all runs newest-first for any run with matching outputs.
func ExtractFromMessages ¶
ExtractFromMessages extracts content from a messages array by role.
func ExtractToolSequence ¶
ExtractToolSequence extracts the sequence of tool names from runs.
func ExtractTraceInputs ¶
ExtractTraceInputs extracts the primary input from a trace's root run.
func ExtractTraceOutput ¶
ExtractTraceOutput extracts the primary output from a trace's root run.
func ExtractValue ¶
func ExtractValue(data map[string]any, fields []string, commonFields []string, messageRole string, fallbackToRaw bool) any
ExtractValue extracts a value from a dict using a priority chain.
func FindRetrievalData ¶
FindRetrievalData finds retrieval data from retriever runs.
Types ¶
type RunData ¶
type RunData struct {
RunID string `json:"run_id"`
TraceID string `json:"trace_id"`
Name string `json:"name"`
RunType string `json:"run_type"`
ParentRunID string `json:"parent_run_id"`
StartTime string `json:"start_time"`
EndTime string `json:"end_time"`
Status string `json:"status"`
Inputs map[string]any `json:"inputs"`
Outputs map[string]any `json:"outputs"`
Error string `json:"error"`
Tags []string `json:"tags"`
}
RunData represents a single run loaded from a trace file.
type Trace ¶
Trace represents a loaded trace: (trace_id, root_run, all_runs).
func LoadTracesFromDir ¶
LoadTracesFromDir loads traces from a directory of JSONL files.
Source Files