INDEX

Explanations

instances where individuals believe they can achieve certain outcomes or accomplish specific tasks

oai_token-act-pair · gpt-3.5-turbo

words indicating the ability or possibility of an action

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

furt

-0.70

 Federation

-0.68

 revision

-0.66

 Likes

-0.63

 Rings

-0.63

 rejection

-0.61

 Irving

-0.61

 Strikes

-0.60

 Yards

-0.59

 rehearsal

-0.58

POSITIVE LOGITS

't

1.60

berra

1.18

adian

1.11

NOT

1.08

 afford

1.07

ieve

0.89

 easily

0.87

isters

0.87

vas

0.86

 regulate

0.83

Activations Density 0.179%

instances where individuals believe they can achieve certain outcomes or accomplish specific tasks

words indicating the ability or possibility of an action

No Comments

No Known Activations