INDEX

Explanations

references to negative social or political situations

oai_token-act-pair · gpt-3.5-turbo

occurrences of the phrase 'on' in various contexts, particularly related to social issues and events

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 11-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.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.11.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

EngineDebug

-0.70

çĶ°

-0.66

peat

-0.65

htar

-0.64

hov

-0.63

enf

-0.62

UME

-0.62

 msec

-0.62

robe

-0.60

[&

-0.59

POSITIVE LOGITS

 Twitter

1.37

 twitter

1.32

 social

1.32

 campuses

1.30

 Facebook

1.23

 blogs

1.22

 facebook

1.18

 forums

1.12

 Reddit

1.11

 behalf

1.09

Activations Density 0.159%

references to negative social or political situations

occurrences of the phrase 'on' in various contexts, particularly related to social issues and events

No Comments

No Known Activations