INDEX

Explanations

No Explanations Found

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

fires

-0.67

 purs

-0.62

 McDonnell

-0.60

 intervening

-0.59

 Clement

-0.58

 Grounds

-0.57

 patched

-0.56

ament

-0.56

weather

-0.56

 Shall

-0.56

POSITIVE LOGITS

 therap

0.94

©¶æ

0.84

 perspect

0.77

phal

0.76

 veter

0.75

 obser

0.73

auga

0.71

 enthusi

0.71

orget

0.71

ilater

0.70

Activations Density 0.000%

No Known Activations

This feature has no known activations.