INDEX

Explanations

phrases discussing challenging, problematic, or contentious issues

oai_token-act-pair · gpt-3.5-turbo

terms related to contentious or thorny issues in discussions

oai_token-act-pair · gpt-4o-mini Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.hook_resid_pre

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

24,576

Data Type

torch.float32

Hook Point

blocks.5.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

uthor

-0.82

ocard

-0.72

 Excellence

-0.72

ucle

-0.71

oreal

-0.71

opath

-0.71

 shelves

-0.70

elsen

-0.69

ugen

-0.68

 stocked

-0.67

POSITIVE LOGITS

 thorn

1.01

 dispute

0.80

 questions

0.78

ifact

0.75

 disputes

0.75

 debated

0.75

tar

0.74

vex

0.73

 debate

0.70

Activations Density 0.048%

phrases discussing challenging, problematic, or contentious issues

terms related to contentious or thorny issues in discussions

No Comments

No Known Activations