INDEX

Explanations

names of politicians or political figures and locations or news agencies

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

iasco

-0.72

udging

-0.70

footed

-0.66

minecraft

-0.64

agy

-0.63

Cyn

-0.61

enhagen

-0.61

aba

-0.60

pread

-0.59

iesta

-0.59

POSITIVE LOGITS

å£«

0.87

roth

0.69

·

0.69

grad

0.68

³

0.68

£

0.68

¶

0.67

Ĥİ

0.67

bilt

0.65

Higher

0.62

Activations Density 0.094%

names of politicians or political figures and locations or news agencies

No Comments

No Known Activations