INDEX

Explanations

words related to geographical locations, specifically focusing on neighborhoods and fault lines

oai_token-act-pair · gpt-3.5-turbo

references to specific neighborhoods and civic-related terms

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.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

etime

-0.80

rael

-0.79

ahime

-0.77

endi

-0.76

tered

-0.76

oiler

-0.76

tering

-0.76

imester

-0.75

undo

-0.74

abl

-0.74

POSITIVE LOGITS

 Pioneer

1.18

 Halls

0.89

 Alpine

0.85

 Beacon

0.82

 Village

0.80

 Cabin

0.79

 Ribbon

0.78

 Echo

0.78

 Galile

0.77

 Trafford

0.74

Activations Density 0.033%

words related to geographical locations, specifically focusing on neighborhoods and fault lines

references to specific neighborhoods and civic-related terms

No Comments

No Known Activations