INDEX

Explanations

phrases related to people recounting information or events

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

isol

-0.79

 sidx

-0.73

 ILCS

-0.70

adesh

-0.69

aband

-0.68

reci

-0.67

agement

-0.65

mun

-0.64

few

-0.64

 squared

-0.63

POSITIVE LOGITS

 reporters

1.17

 HuffPost

0.97

me

0.91

IPS

0.90

us

0.90

 BuzzFeed

0.89

 Politico

0.89

 CNBC

0.87

CNN

0.86

 Newsweek

0.83

Activations Density 0.042%

phrases related to people recounting information or events

No Comments

No Known Activations