diff options
| author | Jonas Knobloch <jonas.knobloch@t-online.de> | 2026-04-24 11:07:01 +0200 |
|---|---|---|
| committer | Jonas Knobloch <jonas.knobloch@t-online.de> | 2026-04-24 11:07:01 +0200 |
| commit | 1e02a5ec7f7794bb542e99ceb804ed88048b6b54 (patch) | |
| tree | 5db849f4eaa326f9b6511c315180e540f00c393e /research | |
| parent | 875ac19922932b12f6357d64a460c06d031000e7 (diff) | |
Access artifacts via shelf
Diffstat (limited to 'research')
| -rw-r--r-- | research/lesci/cmd/lesci/main.go | 9 | ||||
| -rw-r--r-- | research/sander/cmd/sander/main.go | 5 |
2 files changed, 8 insertions, 6 deletions
diff --git a/research/lesci/cmd/lesci/main.go b/research/lesci/cmd/lesci/main.go index ab0a967..c52013a 100644 --- a/research/lesci/cmd/lesci/main.go +++ b/research/lesci/cmd/lesci/main.go @@ -8,6 +8,7 @@ import ( "go.jknobloc.com/x/dataset" "go.jknobloc.com/x/gpt2" "go.jknobloc.com/x/research/lesci" + "go.jknobloc.com/x/shelf" "go.jknobloc.com/x/tokenizer/bpe" ) @@ -40,15 +41,15 @@ func main() { } func setup(control, treatment int) (*lesci.Experiment, *gpt2.Model) { - a := fmt.Sprintf("gpt2/models/onnx_eval/gpt2_%d_m000_babylm_v2", control) - b := fmt.Sprintf("gpt2/models/onnx_eval/gpt2_%d_m000_babylm_v2", 100512) + a := fmt.Sprintf(shelf.Abs("models/mbpe/gpt2_%d_m000_babylm_v2"), control) + b := fmt.Sprintf(shelf.Abs("models/mbpe/gpt2_%d_m000_babylm_v2"), 100512) m := must(model(path.Join(a, "model_eval.onnx"), control)) t := must(bpe.NewTokenizerFromFiles(path.Join(a, "vocab.json"), path.Join(a, "merges.txt"))) c := must(bpe.NewTokenizerFromFiles(path.Join(b, "vocab.json"), path.Join(b, "merges.txt"))) - d := must(dataset.NewFileReader("dataset/cmd/dataset/tmp/babylm/train_100M", "*.train")) + d := must(dataset.NewFileReader(shelf.Abs("data/babylm/train_100M"), "*.train")) - o := fmt.Sprintf("out/lesci/m000/babylm_%d_%d", control, treatment) + o := fmt.Sprintf(shelf.Abs("results/lesci/m000/babylm_%d_%d"), control, treatment) return must(lesci.NewExperiment(m, t, c, d, o, control, 5000)), m } diff --git a/research/sander/cmd/sander/main.go b/research/sander/cmd/sander/main.go index fd9ece3..6f68434 100644 --- a/research/sander/cmd/sander/main.go +++ b/research/sander/cmd/sander/main.go @@ -6,6 +6,7 @@ import ( "path" "go.jknobloc.com/x/research/sander" + "go.jknobloc.com/x/shelf" "go.jknobloc.com/x/tokenizer/bpe" ) @@ -15,8 +16,8 @@ func main() { for _, v := range vocab { for _, a := range alpha { - src := fmt.Sprintf("gpt2/models/mbpe/gpt2_%s_%s_babylm_v2", v, a) - dst := fmt.Sprintf("out/sander/mbpe/gpt2_%s_%s_babylm_v2", v, a) + src := fmt.Sprintf(shelf.Abs("models/mbpe/gpt2_%s_%s_babylm_v2"), v, a) + dst := fmt.Sprintf(shelf.Abs("results/sander/mbpe/gpt2_%s_%s_babylm_v2"), v, a) if err := run(src, dst); err != nil { log.Fatal(err) |
