INDEX

Explanations

phrases related to receiving newsletters or information

oai_token-act-pair · gpt-3.5-turbo

occurrences of the word "Brief."

oai_token-act-pair · gpt-4o-mini Triggered by @bot

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

SEA

-0.80

CE

-0.80

ç¥ŀ

-0.71

OPE

-0.71

JV

-0.69

DJ

-0.69

Drive

-0.69

 technologically

-0.67

University

-0.67

Il

-0.66

POSITIVE LOGITS

 Brief

1.13

cases

1.08

rief

1.02

case

1.00

itude

0.87

gements

0.86

ing

0.85

edly

0.80

 sketches

0.80

gments

0.80

Activations Density 0.007%

phrases related to receiving newsletters or information

occurrences of the word "Brief."

No Comments

No Known Activations