INDEX

Explanations

information related to government actions and policies, particularly concerning health and safety regulations

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

hari

-0.80

KT

-0.78

ording

-0.78

nels

-0.76

obin

-0.75

holes

-0.75

sson

-0.74

lehem

-0.74

anges

-0.72

stad

-0.71

POSITIVE LOGITS

own

1.38

 newest

1.26

 biggest

1.13

 flagship

1.10

 finest

1.05

 namesake

1.04

 inability

1.04

 foray

1.01

 demise

1.01

 predecessor

1.01

Activations Density 1.799%

information related to government actions and policies, particularly concerning health and safety regulations

No Comments

No Known Activations