INDEX

Explanations

phrases related to refuting claims or presenting evidence in a debate or argument

oai_token-act-pair · gpt-3.5-turbo Triggered by @bot

claims related to fabricating or misrepresenting information

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 12-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.hook_resid_post

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_post

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

 tranqu

-0.74

 spirits

-0.73

 solitude

-0.72

 rituals

-0.70

 dreams

-0.69

 nerves

-0.68

anx

-0.68

 goodbye

-0.67

aptic

-0.67

ukong

-0.66

POSITIVE LOGITS

 debunked

1.21

 misrepresent

1.14

 misleading

1.11

 inaccur

1.11

 False

1.10

False

1.09

 inacc

1.08

 FALSE

1.07

Claim

1.07

 falsehood

1.06

Activations Density 0.871%

phrases related to refuting claims or presenting evidence in a debate or argument

claims related to fabricating or misrepresenting information

No Comments

No Known Activations