INDEX

Explanations

phrases related to citing reasons or explanations

oai_token-act-pair · gpt-3.5-turbo

phrases that attribute reasons or justifications, particularly linked to actions or endorsements

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.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

cair

-0.74

mop

-0.69

llular

-0.66

orr

-0.65

medium

-0.65

leaf

-0.65

rams

-0.65

rone

-0.64

roup

-0.63

roman

-0.62

POSITIVE LOGITS

 justification

1.25

 evidence

1.09

 examples

1.08

 reasons

1.07

 inspiration

1.06

 proof

1.03

 contributing

1.02

 reason

1.01

evidence

1.00

 culprit

0.97

Activations Density 0.084%

phrases related to citing reasons or explanations

phrases that attribute reasons or justifications, particularly linked to actions or endorsements

No Comments

No Known Activations