diff options
| -rw-r--r-- | llm/run.go | 16 |
1 files changed, 14 insertions, 2 deletions
@@ -52,13 +52,25 @@ func (e *Evaluator[R]) Run(title string, data dataset.Reader, window, stride int return int(e.completed.Load()) }) - defer pb.Close() + s := 0 // skipped + n := 0 // total - n := 0 + defer func() { + fmt.Printf("skipped %d of %d texts\n", s, n) + }() + + defer pb.Close() for d := range data.Texts("text") { tokens := toInt64(e.tokenizer.Tokenize(d)) + if len(tokens) == 0 { + n++ + s++ + + continue + } + e.schedule(n, tokens, window, stride, e.batchSize) pb.SetTotal(pb.Total() + e.estimateJobs(tokens, window, stride)) |
