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

 franc

-0.72

 initiative

-0.66

 coincide

-0.66

 Catal

-0.64

 Overt

-0.61

LAR

-0.61

 Centre

-0.60

Lub

-0.59

lod

-0.59

cale

-0.58

POSITIVE LOGITS

gettable

0.82

etics

0.72

hett

0.72

rites

0.71

AAAAAAAA

0.71

fters

0.69

ç¥ŀ

0.69

TPPStreamerBot

0.69

essee

0.67

sold

0.67

Activations Density 0.000%

No Known Activations

This feature has no known activations.