INDEX

Explanations

requests or comments related to feedback and interactions

oai_token-act-pair · gpt-3.5-turbo

references to communication, particularly emails and comments

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 theorem

-0.63

 curfew

-0.62

scope

-0.61

unin

-0.60

 autonomy

-0.60

 subsistence

-0.60

osuke

-0.59

 tripod

-0.59

lux

-0.59

 proficiency

-0.59

POSITIVE LOGITS

 thanking

1.01

 poured

1.01

 pouring

0.98

 inbox

0.97

 flooded

0.96

 congrat

0.90

 hurled

0.89

 praising

0.89

 emailed

0.89

 lately

0.88

Activations Density 0.251%

requests or comments related to feedback and interactions

references to communication, particularly emails and comments

No Comments

No Known Activations