INDEX

Explanations

terms related to locations or places, specifically the term "canton."

oai_token-act-pair · gpt-3.5-turbo

repeated mentions of the word "cant."

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 2-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.2.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.2.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

 Generation

-0.84

Ô

-0.81

å£«

-0.69

edience

-0.68

CVE

-0.66

izont

-0.65

issance

-0.65

equality

-0.65

 Flavoring

-0.65

Bang

-0.64

POSITIVE LOGITS

ional

0.82

 dont

0.77

 cant

0.76

peg

0.73

pole

0.73

 toile

0.72

cha

0.72

ember

0.70

avorite

0.69

nery

0.68

Activations Density 0.006%

terms related to locations or places, specifically the term "canton."

repeated mentions of the word "cant."

No Comments

No Known Activations