INDEX

Explanations

phrases related to equality and social justice

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

asus

-0.81

ibur

-0.77

ESE

-0.76

resa

-0.75

edient

-0.72

Ã©e

-0.72

NetMessage

-0.70

 Palestin

-0.69

etts

-0.68

Ã©s

-0.68

POSITIVE LOGITS

 technically

1.24

 admittedly

1.04

 occasional

1.00

 ostensibly

1.00

 slight

0.94

 initially

0.92

 disagree

0.90

 outward

0.87

 caveats

0.87

 setbacks

0.84

Activations Density 2.433%

phrases related to equality and social justice

No Comments

No Known Activations