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

ueller

-0.71

 tweets

-0.67

CHR

-0.65

rontal

-0.65

agi

-0.65

 retweet

-0.64

orah

-0.63

 tweet

-0.60

ãĥĥ

-0.59

 validated

-0.59

POSITIVE LOGITS

Assembly

0.77

Management

0.76

Introdu

0.76

inct

0.69

Prop

0.69

 cumbers

0.69

 Absent

0.67

Ens

0.65

Usage

0.65

ggles

0.64

Activations Density 0.000%

No Known Activations

This feature has no known activations.