INDEX

Explanations

phrases related to achievements and success

oai_token-act-pair · gpt-3.5-turbo

references to accomplishments or successes

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 4-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.4.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.4.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

ondo

-0.80

sey

-0.73

cos

-0.69

cise

-0.68

ocard

-0.68

avis

-0.67

creen

-0.67

©¶æ

-0.66

odynamic

-0.64

avin

-0.63

POSITIVE LOGITS

 achievement

1.14

 accomplishment

1.10

 achie

1.07

 achievements

1.04

 attainment

1.02

 Achievement

0.97

 accomplished

0.96

 feats

0.94

rious

0.88

ment

0.87

Activations Density 0.018%

phrases related to achievements and success

references to accomplishments or successes

No Comments

No Known Activations