INDEX

Explanations

email content-related phrases

oai_token-act-pair · gpt-3.5-turbo

terms related to content and its availability

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.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

rolet

-0.78

STER

-0.70

Äĩ

-0.69

STON

-0.68

 Siem

-0.68

 Sons

-0.67

 induct

-0.66

STRUCT

-0.64

ESCO

-0.63

Â¯Â¯Â¯Â¯Â¯Â¯Â¯Â¯Â¯Â¯Â¯Â¯Â¯Â¯Â¯Â¯

-0.63

POSITIVE LOGITS

edly

1.19

Content

1.03

 content

0.91

 Content

0.86

content

0.83

ais

0.75

fill

0.72

ucha

0.72

skin

0.71

illary

0.69

Activations Density 0.009%

email content-related phrases

terms related to content and its availability

No Comments

No Known Activations