INDEX

Explanations

numbers or numeric information

oai_token-act-pair · gpt-3.5-turbo

numerical data related to viewers and ratings of shows

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 11-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.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.11.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

bard

-0.86

Alz

-0.80

 Hast

-0.73

 healer

-0.71

 Shap

-0.70

 gard

-0.70

 Alchemist

-0.70

Khe

-0.69

 Plague

-0.67

raft

-0.67

POSITIVE LOGITS

 viewership

1.66

 ratings

1.42

 viewers

1.41

 Nielsen

1.37

 viewer

1.19

 rating

1.17

 audience

1.14

 Ratings

1.11

 cable

1.11

 audiences

1.09

Activations Density 0.135%

numbers or numeric information

numerical data related to viewers and ratings of shows

No Comments

No Known Activations