INDEX

Explanations

phrases related to political and legal issues

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

 undertaking

-0.65

-0.62

nesty

-0.61

ignt

-0.61

Hai

-0.59

COMPLE

-0.57

cellence

-0.57

watching

-0.57

 issu

-0.56

ikhail

-0.56

POSITIVE LOGITS

fully

1.14

 sparing

1.02

FUL

1.02

 techniques

0.89

fulness

0.89

ful

0.87

 aliases

0.86

 condoms

0.86

 tools

0.83

 shortcuts

0.81

Activations Density 3.795%

phrases related to political and legal issues

No Comments

No Known Activations