INDEX

Explanations

words related to being filled or abundant with something, often with a negative connotation

oai_token-act-pair · gpt-3.5-turbo

terms indicating problematic or difficult situations

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

ift

-0.72

mpeg

-0.72

Balt

-0.70

intestinal

-0.68

hal

-0.66

Ku

-0.66

isting

-0.66

 axis

-0.63

ifted

-0.63

friend

-0.63

POSITIVE LOGITS

entimes

0.86

arella

0.74

 rife

0.73

odies

0.72

oche

0.70

ttes

0.70

 Spartan

0.67

 littered

0.67

geries

0.67

 riddled

0.65

Activations Density 0.032%

words related to being filled or abundant with something, often with a negative connotation

terms indicating problematic or difficult situations

No Comments

No Known Activations