INDEX

Explanations

phrases or numbers representing 'two-way' interactions or comparisons

oai_token-act-pair · gpt-3.5-turbo

phrases indicating a two-party system or duality in 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

Ces

-0.67

 Rodrigo

-0.64

 imagination

-0.63

 Gillespie

-0.62

 commentary

-0.61

URRENT

-0.59

olics

-0.59

Gru

-0.59

 Canaver

-0.59

 Tuls

-0.59

POSITIVE LOGITS

thirds

1.86

dimensional

1.56

legged

1.52

hander

1.43

sided

1.41

dozen

1.38

way

1.36

tier

1.35

faced

1.34

bedroom

1.33

Activations Density 0.038%

phrases or numbers representing 'two-way' interactions or comparisons

phrases indicating a two-party system or duality in contexts

No Comments

No Known Activations