INDEX

Explanations

mentions of forceful criticism or disapproval

oai_token-act-pair · gpt-3.5-turbo

occurrences of the word "slammed" and its variants in various contexts

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

vernment

-0.80

abet

-0.76

vous

-0.72

 Carbuncle

-0.72

thus

-0.71

nown

-0.70

mental

-0.70

earances

-0.69

 Sight

-0.68

chn

-0.67

POSITIVE LOGITS

 slammed

1.19

 slamming

1.07

 slams

1.02

 slam

0.91

 dunk

0.90

 hitters

0.84

 shut

0.82

 heels

0.81

 smack

0.81

 elbows

0.78

Activations Density 0.012%

mentions of forceful criticism or disapproval

occurrences of the word "slammed" and its variants in various contexts

No Comments

No Known Activations