INDEX

Explanations

words related to negative impacts or flaws

oai_token-act-pair · gpt-3.5-turbo

words and phrases related to damage or blemishes on reputation or appearance

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

agy

-0.78

WER

-0.76

yip

-0.76

raltar

-0.75

cffffcc

-0.73

trak

-0.72

ultane

-0.72

ultan

-0.71

 guiActiveUn

-0.71

 ãĤµãĥ¼ãĥĨãĤ£ãĥ¯ãĥ³

-0.70

POSITIVE LOGITS

 tarn

1.02

 stains

0.91

 stain

0.87

luster

0.82

 stained

0.82

 coating

0.82

 smear

0.81

uous

0.80

 linen

0.77

 washed

0.77

Activations Density 0.167%

words related to negative impacts or flaws

words and phrases related to damage or blemishes on reputation or appearance

No Comments

No Known Activations