INDEX

Explanations

political and authoritarian-themed terms and actions

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_fs768-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_768

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

768

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

éĹĺ

-0.73

kHz

-0.69

hower

-0.67

terday

-0.66

ignty

-0.66

 apprehension

-0.62

cob

-0.61

kHz

-0.60

Ae

-0.60

 Siem

-0.59

POSITIVE LOGITS

gers

1.39

glers

1.31

rett

1.21

mented

1.13

rant

1.12

ging

1.11

gy

1.08

mentation

1.08

hett

1.05

herer

1.04

Activations Density 4.489%

political and authoritarian-themed terms and actions

No Comments

No Known Activations