INDEX

Explanations

word sequences indicating a statement or argument is being made

oai_token-act-pair · gpt-3.5-turbo

statements of opinion or argumentation

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

ractor

-0.80

ioxide

-0.77

wow

-0.77

MpServer

-0.75

incial

-0.75

ogyn

-0.75

scribe

-0.73

crow

-0.72

Mine

-0.71

obyl

-0.69

POSITIVE LOGITS

 eliminating

1.22

 removing

1.18

 allowing

1.15

 preserving

1.15

 abandoning

1.13

 delaying

1.13

 restricting

1.13

 restoring

1.12

 maintaining

1.12

 limiting

1.12

Activations Density 0.316%

word sequences indicating a statement or argument is being made

statements of opinion or argumentation

No Comments

No Known Activations