INDEX

Explanations

words related to intelligence and judgment (e.g., 'stupid', 'dumb', 'smart')

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

riott

-0.81

AUT

-0.79

accompan

-0.76

APH

-0.76

apers

-0.75

ILA

-0.74

orthy

-0.73

Reviewed

-0.70

OHN

-0.68

-0.66

POSITIVE LOGITS

founded

1.16

found

0.97

nesses

0.90

ness

0.90

fuck

0.89

ly

0.88

est

0.87

asses

0.85

itude

0.84

stru

0.84

Activations Density 0.036%

words related to intelligence and judgment (e.g., 'stupid', 'dumb', 'smart')

No Comments

No Known Activations