INDEX

Explanations

words related to instructions or actions that involve visiting websites or engaging with online content

oai_token-act-pair · gpt-3.5-turbo

the word "to" indicating actions or intentions

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 9-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.9.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.9.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

ELD

-0.70

offensive

-0.67

 trailing

-0.65

standing

-0.63

 Said

-0.63

 layoffs

-0.61

existent

-0.61

enforcement

-0.59

soDeliveryDate

-0.57

soType

-0.56

POSITIVE LOGITS

 learn

1.32

see

1.21

 find

1.20

get

1.20

 avoid

1.18

 maximize

1.15

ggles

1.15

 obtain

1.13

 receive

1.10

 ensure

1.10

Activations Density 0.139%

words related to instructions or actions that involve visiting websites or engaging with online content

the word "to" indicating actions or intentions

No Comments

No Known Activations