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

fam

-0.83

reshold

-0.80

iffe

-0.80

fp

-0.75

ACC

-0.75

utter

-0.73

rha

-0.71

imp

-0.68

CF

-0.68

antage

-0.67

POSITIVE LOGITS

 spotting

0.75

ãĥ¬

0.73

 infiltration

0.72

ndra

0.70

 advising

0.67

 hostilities

0.67

 Sharing

0.66

Sense

0.65

 Lighting

0.64

 diplomacy

0.64

Activations Density 0.000%

No Known Activations

This feature has no known activations.