From 408536516bdfedc93d93226dc08280ed224361fc Mon Sep 17 00:00:00 2001 From: Jonas Knobloch Date: Fri, 24 Apr 2026 11:14:43 +0200 Subject: Cleanup eval command * Remove logprob extraction --- llm/cmd/eval/ppl.go | 78 ----------------------------------------------------- 1 file changed, 78 deletions(-) delete mode 100644 llm/cmd/eval/ppl.go (limited to 'llm/cmd/eval/ppl.go') diff --git a/llm/cmd/eval/ppl.go b/llm/cmd/eval/ppl.go deleted file mode 100644 index 2394d83..0000000 --- a/llm/cmd/eval/ppl.go +++ /dev/null @@ -1,78 +0,0 @@ -package main - -import ( - "fmt" - "log" - "math" - "strings" - - "go.jknobloc.com/x/dataset" - "go.jknobloc.com/x/llm" -) - -type pplResult struct { - v float64 - n int -} - -func perplexity() { - d := data() - - m := model() - t := tokenizer() - - e := llm.NewEvaluator(m, t, func(job llm.Job, logProbs []float32, tokens []int) pplResult { - total := float64(0) - - n := 0 - - for _, p := range logProbs { - total -= float64(p) - - n++ - } - - return pplResult{ - v: total, - n: n, - } - }, llm.EvaluatorConfig{ - BatchSize: 1, - NumWorkers: 4, - }) - - total := float64(0) - n := 0 - - if err := e.RunAndCollect("Perplexity", d, 1024, 512, func(r pplResult) error { - total += r.v - n += r.n - - return nil - }); err != nil { - log.Fatal(err) - } - - avg := total / float64(n) - ppl := math.Exp(avg) - - fmt.Println(ppl) - - if err := m.Destroy(); err != nil { - log.Fatal(err) - } -} - -func joined() dataset.Reader { - miniPile := data() - - docs := make([]string, 0) - - for _, d := range miniPile.Texts() { - docs = append(docs, d) - } - - j := dataset.NewStringReader(strings.Join(docs, "\n\n")) - - return j -} -- cgit v1.3.1