INDEX

Explanations

words related to visual or sensory evaluation such as "looks", "tastes", "feels"

oai_token-act-pair · gpt-3.5-turbo

descriptions of visual appearances

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.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

limit

-0.73

ulla

-0.69

learning

-0.67

ilings

-0.67

upuncture

-0.67

mental

-0.66

trl

-0.66

Osw

-0.65

venient

-0.64

ference

-0.63

POSITIVE LOGITS

 suspic

0.93

 like

0.89

 awfully

0.88

 identical

0.84

 strikingly

0.83

 blurry

0.81

 sleek

0.80

 vaguely

0.79

 prett

0.79

 shiny

0.78

Activations Density 0.068%

words related to visual or sensory evaluation such as "looks", "tastes", "feels"

descriptions of visual appearances

No Comments

No Known Activations