INDEX

Explanations

numeric patterns in the format "number-number" indicating a range of values

oai_token-act-pair · gpt-3.5-turbo

phrases related to numerical scales or ratings

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 7-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.7.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.7.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

 Mechdragon

-0.69

 cheated

-0.65

 solicitation

-0.64

SIG

-0.63

lax

-0.59

 Canary

-0.58

ãģ®å®

-0.58

BCC

-0.58

 Rolls

-0.57

 Toro

-0.57

POSITIVE LOGITS

hour

0.88

four

0.83

month

0.83

three

0.83

seven

0.83

hours

0.78

equal

0.78

quart

0.78

abit

0.77

poons

0.77

Activations Density 0.068%

numeric patterns in the format "number-number" indicating a range of values

phrases related to numerical scales or ratings

No Comments

No Known Activations