INDEX

Explanations

references to the Fox News network

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_fs12288-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_12288

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

12,288

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

rians

-0.82

rian

-0.74

ansom

-0.72

acterial

-0.71

akeru

-0.71

 inval

-0.70

ictions

-0.66

assy

-0.65

orically

-0.65

rella

-0.64

POSITIVE LOGITS

conn

1.49

 News

0.92

hawk

0.87

News

0.87

woods

0.83

fire

0.81

 Broadcasting

0.81

cat

0.79

Wire

0.77

croft

0.77

Activations Density 0.050%

references to the Fox News network

No Comments

No Known Activations