INDEX

Explanations

time durations specified in hours within a text

oai_token-act-pair · gpt-3.5-turbo

references to time periods, particularly those indicating a duration of 24 hours

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

yip

-0.81

olini

-0.71

oslav

-0.65

agogue

-0.63

 Sisters

-0.63

hammad

-0.63

yden

-0.62

orically

-0.62

ãĤ©

-0.59

 Swanson

-0.58

POSITIVE LOGITS

hour

0.99

hours

0.87

0.86

Hour

0.85

0.82

0.74

kar

0.74

hack

0.70

seven

0.68

Hours

0.66

Activations Density 0.084%

time durations specified in hours within a text

references to time periods, particularly those indicating a duration of 24 hours

No Comments

No Known Activations