INDEX

Explanations

statements and actions related to pressure, influence, and intimidation within various contexts

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res_fs24576-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_24576

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

24,576

Data Type

torch.float32

Hook Point

blocks.8.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

wayne

-0.79

tis

-0.77

alg

-0.73

iasm

-0.70

}}}

-0.69

!.

-0.68

aho

-0.67

gae

-0.66

();

-0.65

folk

-0.64

POSITIVE LOGITS

 inappropriately

1.07

 inappropriate

0.87

 Paddock

0.85

 unlawfully

0.82

 improper

0.80

 improperly

0.79

 prostitutes

0.78

 deliberately

0.77

"'

0.77

 Blumenthal

0.77

Activations Density 0.441%

statements and actions related to pressure, influence, and intimidation within various contexts

No Comments

No Known Activations