INDEX

Explanations

proper names of individuals

oai_token-act-pair · gpt-3.5-turbo

names of individuals and instances of derogatory or disparaging language

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

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

Tok

-0.82

aido

-0.72

bos

-0.71

 replication

-0.70

Tok

-0.70

rawdownloadcloneembedreportprint

-0.68

slave

-0.68

Phase

-0.68

BF

-0.68

 injection

-0.68

POSITIVE LOGITS

 Moreno

2.13

 Hayden

2.04

 derogatory

1.58

 dispar

1.32

 depl

1.26

 Flat

1.24

 Chin

1.04

 Jerome

0.97

 Fern

0.94

 Chester

0.94

Activations Density 0.042%

proper names of individuals

names of individuals and instances of derogatory or disparaging language

No Comments

No Known Activations