INDEX

Explanations

phrases related to negative impacts or consequences

oai_token-act-pair · gpt-3.5-turbo

occurrences of the phrase "taking a toll" or similar expressions that indicate negative consequences

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

 Recomm

-0.78

 Cheong

-0.71

urrent

-0.70

displayText

-0.66

tions

-0.65

zu

-0.64

ado

-0.63

ossier

-0.62

 Archangel

-0.62

assies

-0.61

POSITIVE LOGITS

 toll

1.39

 beating

0.85

 aback

0.85

 Toll

0.84

 tumble

0.84

 notor

0.82

 pounding

0.81

 shine

0.78

 shape

0.77

 twist

0.74

Activations Density 0.080%

phrases related to negative impacts or consequences

occurrences of the phrase "taking a toll" or similar expressions that indicate negative consequences

No Comments

No Known Activations