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

ķ

-0.70

raft

-0.69

 sonic

-0.66

Ļ

-0.65

©

-0.65

 Brand

-0.65

µ

-0.63

 Tier

-0.63

 independence

-0.63

 Codec

-0.63

POSITIVE LOGITS

jri

1.06

jee

0.87

prus

0.80

etsk

0.80

 Scholars

0.79

urious

0.78

rador

0.78

phis

0.77

irez

0.75

jriwal

0.75

Activations Density 0.000%

No Known Activations

This feature has no known activations.