INDEX

Explanations

words related to political discussions or controversies

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_fs1536-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_1536

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

1,536

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

ses

-0.77

ãģĨ

-0.74

-0.58

())

-0.58

thro

-0.57

url

-0.57

recomm

-0.57

pt

-0.57

://

-0.56

fell

-0.56

POSITIVE LOGITS

romeda

1.20

rea

0.98

 therein

0.90

 secondly

0.85

hra

0.85

ersen

0.82

 furthermore

0.81

 lest

0.81

 besides

0.81

fter

0.80

Activations Density 1.477%

words related to political discussions or controversies

No Comments

No Known Activations