INDEX

Explanations

instances where something has been observed or noticed

oai_token-act-pair · gpt-3.5-turbo

instances of the word "seen" used to convey observations or experiences

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.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

ÃŃa

-0.76

prime

-0.72

vernment

-0.72

atto

-0.71

 contrace

-0.69

trap

-0.69

angan

-0.69

plan

-0.62

assisted

-0.61

 Ã©

-0.60

POSITIVE LOGITS

dust

0.86

ĸļ

0.82

 Seen

0.79

peak

0.77

Ĺ

0.77

seeing

0.72

PsyNetMessage

0.69

 seen

0.66

 glimps

0.66

icia

0.66

Activations Density 0.032%

instances where something has been observed or noticed

instances of the word "seen" used to convey observations or experiences

No Comments

No Known Activations