INDEX

Explanations

No Explanations Found

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.hook_resid_pre

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

24,576

Data Type

torch.float32

Hook Point

blocks.0.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 wounding

-0.68

 prof

-0.64

 arts

-0.64

 stabbing

-0.64

DEBUG

-0.63

 relaxation

-0.62

 bystand

-0.61

 prom

-0.60

fo

-0.60

lo

-0.59

POSITIVE LOGITS

Registered

0.77

å§«

0.75

igible

0.75

Mov

0.74

 Perkins

0.74

ilater

0.73

answer

0.72

 Rowling

0.71

abwe

0.70

 Granger

0.70

Activations Density 0.000%

No Known Activations

This feature has no known activations.