INDEX

Explanations

phrases containing the word "presses."

oai_token-act-pair · gpt-3.5-turbo

individual words or terms that have a high frequency of occurrence

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 advoc

-0.67

spread

-0.66

Ü

-0.63

ilater

-0.63

Ïī

-0.62

Ò

-0.62

 predec

-0.62

Ĵ

-0.61

 tradem

-0.60

 suspic

-0.59

POSITIVE LOGITS

 Profile

0.92

 meanwhile

0.76

 Companion

0.69

 Edit

0.68

 strives

0.68

 Brothers

0.66

 Pharmaceutical

0.65

 disagrees

0.64

 publishes

0.63

 denies

0.62

Activations Density 0.659%

phrases containing the word "presses."

individual words or terms that have a high frequency of occurrence

No Comments

No Known Activations