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

 theat

-0.71

GM

-0.67

ãĥŀ

-0.66

 ACTIONS

-0.65

OUGH

-0.62

terday

-0.61

onomy

-0.61

Â¯Â¯Â¯Â¯

-0.61

âĸº

-0.61

Josh

-0.60

POSITIVE LOGITS

steen

0.76

verages

0.73

mite

0.67

 Aval

0.67

 Junk

0.66

zi

0.63

anners

0.63

Veg

0.63

puff

0.63

ql

0.63

Activations Density 0.000%

No Known Activations

This feature has no known activations.