INDEX

Explanations

proper nouns or names, specifically ending with 'ie'

oai_token-act-pair · gpt-3.5-turbo

occurrences of the word "ie."

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

è¦ļéĨĴ

-0.87

 destro

-0.78

 glim

-0.76

 upkeep

-0.76

 gobl

-0.73

ŃĶ

-0.73

ipeg

-0.73

 disemb

-0.72

pmwiki

-0.72

ickr

-0.72

POSITIVE LOGITS

gie

1.04

ie

0.99

lla

0.93

gee

0.92

Mae

0.90

jay

0.86

llo

0.85

ux

0.84

jee

0.83

bies

0.81

Activations Density 0.017%

proper nouns or names, specifically ending with 'ie'

occurrences of the word "ie."

No Comments

No Known Activations