INDEX

Explanations

negative attributes or events

oai_token-act-pair · gpt-3.5-turbo

negative connotations or references to bad situations

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.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

ĸļ

-0.90

racuse

-0.74

ittees

-0.74

 Lauder

-0.71

olate

-0.67

illon

-0.67

inguished

-0.66

 Pavilion

-0.65

theless

-0.64

 Pione

-0.64

POSITIVE LOGITS

dies

1.43

gered

1.29

die

1.19

ger

1.19

dest

1.10

asses

1.05

 luck

1.03

gers

1.03

der

0.98

 publicity

0.97

Activations Density 0.038%

negative attributes or events

negative connotations or references to bad situations

No Comments

No Known Activations