INDEX

Explanations

mentions of specific names, likely related to Twitter accounts or organizations

oai_token-act-pair · gpt-3.5-turbo

instances of proper nouns and names, particularly associated with people and organizations

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

 Lore

-0.94

 Texans

-0.92

 Kard

-0.85

Rx

-0.85

oy

-0.84

TAG

-0.84

OW

-0.83

 Ring

-0.83

PW

-0.82

Soy

-0.80

POSITIVE LOGITS

Ha

1.03

 Churchill

0.98

Cho

0.96

 Brighton

0.93

 Devil

0.93

Ha

0.90

 Chaos

0.89

 Haas

0.89

 Hastings

0.86

 Devils

0.86

Activations Density 0.491%

mentions of specific names, likely related to Twitter accounts or organizations

instances of proper nouns and names, particularly associated with people and organizations

No Comments

No Known Activations