INDEX

Explanations

phrases referencing or mentioning something or someone

oai_token-act-pair · gpt-3.5-turbo

instances of the word "referring" and similar phrases that indicate attribution or citation of information

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 4-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.4.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.4.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

 foothold

-0.69

hold

-0.68

tes

-0.66

tmp

-0.65

mega

-0.64

»Ĵ

-0.62

rieve

-0.61

sburgh

-0.60

eu

-0.60

 lineback

-0.60

POSITIVE LOGITS

 sarcast

0.89

 referring

0.88

 sarc

0.85

Nib

0.75

Jub

0.73

 quoting

0.71

 joking

0.69

 insult

0.69

ento

0.68

 homophobic

0.68

Activations Density 0.056%

phrases referencing or mentioning something or someone

instances of the word "referring" and similar phrases that indicate attribution or citation of information

No Comments

No Known Activations