INDEX

Explanations

references to specific individuals and pronouns associated with actions towards those individuals

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

 Magazine

-0.66

iott

-0.65

icion

-0.64

inelli

-0.63

Magazine

-0.63

Comment

-0.63

 Seller

-0.60

plays

-0.60

icians

-0.59

cean

-0.59

POSITIVE LOGITS

 captive

0.95

 alive

0.89

 away

0.83

 willingly

0.82

 prisoner

0.81

 into

0.80

 accountable

0.80

 afloat

0.79

vati

0.79

 unconscious

0.79

Activations Density 0.122%

references to specific individuals and pronouns associated with actions towards those individuals

No Comments

No Known Activations