INDEX

Explanations

numeric ratings in the form of stars

oai_token-act-pair · gpt-3.5-turbo

references to star ratings 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 @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

ĸļ

-0.90

£ı

-0.89

 Canaver

-0.89

odcast

-0.86

pse

-0.77

ptives

-0.75

nown

-0.73

apons

-0.70

hod

-0.69

sym

-0.69

POSITIVE LOGITS

vation

0.99

rating

0.93

rated

0.84

ved

0.80

Rated

0.78

 ratings

0.76

 rating

0.75

ochet

0.72

bucks

0.72

star

0.72

Activations Density 0.013%

numeric ratings in the form of stars

references to star ratings in various contexts

No Comments

No Known Activations