INDEX

Explanations

words related to negative attributes or consequences

oai_token-act-pair · gpt-3.5-turbo

instances of the word "ill" and its variations, indicating a focus on concepts relating to negative health or logical fallacies

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.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

ļéĨĴ

-0.75

*/(

-0.75

EStream

-0.74

uyomi

-0.73

kefeller

-0.73

âĹ¼

-0.67

 compr

-0.67

©¶æ¥µ

-0.66

 derog

-0.65

EStreamFrame

-0.64

POSITIVE LOGITS

uminati

1.30

inois

1.12

ogical

1.11

umin

1.08

awar

1.08

iberal

1.06

igan

0.99

nesses

0.98

ison

0.98

usive

0.96

Activations Density 0.009%

words related to negative attributes or consequences

instances of the word "ill" and its variations, indicating a focus on concepts relating to negative health or logical fallacies

No Comments

No Known Activations