INDEX

Explanations

actions or decisions that are seen as mistakes or are associated with wrongdoing

oai_token-act-pair · gpt-3.5-turbo

actions or decisions that indicate wrongdoing or mistakes

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 11-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.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.11.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 periodic

-0.68

assian

-0.65

 presses

-0.62

enser

-0.62

 ageing

-0.60

 traveller

-0.59

rouse

-0.58

ilus

-0.57

 resumes

-0.56

 refill

-0.56

POSITIVE LOGITS

 yesterday

1.25

 last

1.15

 earlier

1.01

 LAST

0.94

ago

0.90

wrong

0.87

didn

0.86

terday

0.85

 wrong

0.84

last

0.82

Activations Density 0.548%

actions or decisions that are seen as mistakes or are associated with wrongdoing

actions or decisions that indicate wrongdoing or mistakes

No Comments

No Known Activations