INDEX

Explanations

phrases indicating a call to action, specifically prompting to perform a specific action or asking for a response

oai_token-act-pair · gpt-3.5-turbo

repeated phrases emphasizing possession or inquiries in a conversational context

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.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

Â¨

-0.67

Bridge

-0.63

 Subcommittee

-0.60

 salute

-0.58

Hum

-0.58

aus

-0.57

 fray

-0.57

 Voters

-0.57

thank

-0.57

osa

-0.56

POSITIVE LOGITS

 been

1.09

been

1.07

 gotten

0.91

 done

0.85

 travelled

0.83

 undergone

0.81

 eaten

0.80

 Been

0.79

 chosen

0.79

 fallen

0.78

Activations Density 0.139%

phrases indicating a call to action, specifically prompting to perform a specific action or asking for a response

repeated phrases emphasizing possession or inquiries in a conversational context

No Comments

No Known Activations