INDEX

Explanations

terms related to direct actions or relationships

oai_token-act-pair · gpt-3.5-turbo

instances of the word "direct" indicating various forms of direct interaction or action

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

 Norn

-0.76

 Cursed

-0.72

Peb

-0.71

nesota

-0.69

abies

-0.69

 Mell

-0.67

 Face

-0.67

 Alive

-0.65

 Disabled

-0.65

 Petersburg

-0.64

POSITIVE LOGITS

 direct

1.11

 indirect

1.01

 sunlight

0.90

irect

0.79

forward

0.76

orial

0.75

htaking

0.75

direct

0.75

cuts

0.74

 distribut

0.74

Activations Density 0.010%

terms related to direct actions or relationships

instances of the word "direct" indicating various forms of direct interaction or action

No Comments

No Known Activations