INDEX

Explanations

No Explanations Found

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.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

erences

-0.65

ãĥ¼ãĥ³

-0.64

 island

-0.63

ande

-0.59

upp

-0.59

 Swap

-0.58

 environment

-0.58

Pul

-0.58

 Luigi

-0.57

 Magnetic

-0.57

POSITIVE LOGITS

NF

0.96

isSpecialOrderable

0.76

RESULTS

0.75

lyak

0.74

pai

0.74

 Baghd

0.74

LOG

0.74

 challeng

0.73

 participated

0.69

inis

0.69

Activations Density 0.000%

No Known Activations

This feature has no known activations.