INDEX

Explanations

adjectives describing a high level or prevalence of something negative or undesirable

oai_token-act-pair · gpt-3.5-turbo

terms related to widespread issues or problems

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

tein

-0.76

ift

-0.73

izons

-0.72

pta

-0.68

rieve

-0.68

agine

-0.66

thood

-0.65

bands

-0.65

aiden

-0.65

ARY

-0.65

POSITIVE LOGITS

 rampant

0.87

entimes

0.83

 misinformation

0.75

 plagiar

0.74

 counterfe

0.72

uploads

0.70

iously

0.70

 rife

0.70

 prolifer

0.69

ãĥ£

0.67

Activations Density 0.049%

adjectives describing a high level or prevalence of something negative or undesirable

terms related to widespread issues or problems

No Comments

No Known Activations