INDEX

Explanations

negative health-related phrases

oai_token-act-pair · gpt-3.5-turbo

statements related to health concerns and warnings

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.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

endix

-0.75

Finally

-0.75

 Finally

-0.72

cellaneous

-0.71

âķĲ

-0.71

"}

-0.68

))))

-0.67

CONCLUS

-0.67

etheless

-0.66

NetMessage

-0.65

POSITIVE LOGITS

 usually

0.67

 rightfully

0.65

 mainly

0.60

 typically

0.60

 commonly

0.59

 rightly

0.58

Firstly

0.58

 famously

0.58

 traditionally

0.57

 initially

0.57

Activations Density 1.163%

negative health-related phrases

statements related to health concerns and warnings

No Comments

No Known Activations