INDEX

Explanations

words related to negative judgment or criticism

oai_token-act-pair · gpt-3.5-turbo

terms related to condescension and admonishment

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

å£«

-0.89

INESS

-0.77

FUL

-0.71

hare

-0.70

edom

-0.67

 Grail

-0.66

 Ø§ÙĦ

-0.65

crop

-0.65

Jub

-0.65

 Fifth

-0.65

POSITIVE LOGITS

ension

1.10

uated

1.08

uation

1.07

ple

1.05

uating

0.98

ensions

0.96

uations

0.94

uates

0.91

ulum

0.90

 insin

0.90

Activations Density 0.022%

words related to negative judgment or criticism

terms related to condescension and admonishment

No Comments

No Known Activations