INDEX

Explanations

words related to the trustworthiness, validity, or credibility of information or statements

oai_token-act-pair · gpt-3.5-turbo 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

GROUP

-0.74

artifacts

-0.69

 ALSO

-0.69

gdala

-0.68

 abducted

-0.68

 Alive

-0.67

 Surv

-0.64

 Allaah

-0.64

Downloadha

-0.64

=~

-0.62

POSITIVE LOGITS

est

1.42

ly

1.41

hearted

1.24

ness

1.19

ingly

1.19

ening

1.14

headed

1.08

nesses

1.05

liness

1.05

ener

1.03

Activations Density 0.170%

words related to the trustworthiness, validity, or credibility of information or statements

No Comments

No Known Activations