INDEX

Explanations

the word "is" at the end of a statement

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

 resorts

-0.66

 smear

-0.62

sie

-0.61

 shifts

-0.59

igraph

-0.56

itures

-0.55

onym

-0.55

 lash

-0.55

eur

-0.54

 settlements

-0.54

POSITIVE LOGITS

 happening

1.04

 gonna

0.93

 transpired

0.88

 happened

0.87

omething

0.86

 happ

0.81

wered

0.77

 bothering

0.75

leep

0.75

actly

0.74

Activations Density 0.035%

the word "is" at the end of a statement

No Comments

No Known Activations