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

Topics

-0.73

viously

-0.68

itbart

-0.67

ITH

-0.67

}}}

-0.67

".

-0.65

 BART

-0.65

 wiser

-0.64

=/

-0.64

pict

-0.64

POSITIVE LOGITS

outhern

0.87

edo

0.82

keye

0.75

lege

0.70

odium

0.70

ugu

0.69

 Alpine

0.68

auga

0.66

unta

0.66

 wheelchair

0.66

Activations Density 0.000%

No Known Activations

This feature has no known activations.