INDEX

Explanations

parenthetical expressions containing a relatively high level of intensity or importance

oai_token-act-pair · gpt-3.5-turbo

parentheses used for clarification or elaboration

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.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

 millenn

-0.79

 Lumpur

-0.74

quished

-0.73

 lull

-0.70

 inund

-0.69

 therap

-0.68

 nutrient

-0.68

ris

-0.68

 wildlife

-0.66

 intensive

-0.66

POSITIVE LOGITS

sic

1.45

possibly

1.36

albeit

1.35

emphasis

1.34

although

1.33

including

1.32

usually

1.30

perhaps

1.30

which

1.27

see

1.26

Activations Density 0.236%

parenthetical expressions containing a relatively high level of intensity or importance

parentheses used for clarification or elaboration

No Comments

No Known Activations