INDEX

Explanations

hyphenated phrases, especially those containing a negative connotation

oai_token-act-pair · gpt-3.5-turbo

negative terms or phrases indicating criticism or complaint

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 6-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.6.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.6.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

 opin

-0.79

 reflection

-0.72

 incub

-0.70

 sockets

-0.67

 recomp

-0.66

 circumcised

-0.66

 drinks

-0.66

 defic

-0.65

ulhu

-0.65

 sexes

-0.65

POSITIVE LOGITS

style

1.22

esque

1.15

sama

1.15

inspired

1.14

san

1.12

based

1.12

derived

1.11

themed

1.09

winning

1.08

centered

1.05

Activations Density 0.095%

hyphenated phrases, especially those containing a negative connotation

negative terms or phrases indicating criticism or complaint

No Comments

No Known Activations