INDEX

Explanations

phrases related to performance evaluations, particularly emphasizing strong or weak performance

oai_token-act-pair · gpt-3.5-turbo

adjectives describing performance or quality

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

Sources

-0.82

conom

-0.76

skirts

-0.74

galitarian

-0.72

igmat

-0.72

zos

-0.71

xual

-0.68

soDeliveryDate

-0.67

mr

-0.67

 regrets

-0.67

POSITIVE LOGITS

 outing

1.36

 performance

1.18

 finish

1.13

 showing

1.12

 performances

1.04

 debut

1.03

 matchup

1.01

 season

1.00

 sophomore

0.94

 postseason

0.93

Activations Density 0.168%

phrases related to performance evaluations, particularly emphasizing strong or weak performance

adjectives describing performance or quality

No Comments

No Known Activations