INDEX

Explanations

mentions of opposing or contrasting viewpoints

oai_token-act-pair · gpt-3.5-turbo

mentions of "counter" in various contexts

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

OOD

-0.65

 incom

-0.64

 Taste

-0.61

Bib

-0.61

 Slime

-0.60

 goodbye

-0.59

 Fedora

-0.58

ELS

-0.58

 Cotton

-0.58

 Crossref

-0.58

POSITIVE LOGITS

measures

1.48

intuitive

1.47

culture

1.31

fact

1.30

attack

1.30

intelligence

1.30

balance

1.29

cultural

1.24

balanced

1.23

clock

1.22

Activations Density 0.029%

mentions of opposing or contrasting viewpoints

mentions of "counter" in various contexts

No Comments

No Known Activations