INDEX

Explanations

superlatives, such as "most", in a given context

oai_token-act-pair · gpt-3.5-turbo

superlative terms indicating the highest degree or intensity of a characteristic

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

 Tracks

-0.79

zek

-0.71

ortunately

-0.71

Þ

-0.68

 Seym

-0.67

tif

-0.65

dd

-0.64

ategories

-0.64

ijk

-0.64

aven

-0.63

POSITIVE LOGITS

liest

0.97

 possible

0.93

hest

0.86

 amount

0.84

 conceivable

0.78

 practicable

0.78

 bang

0.76

iest

0.75

igators

0.75

 imaginable

0.74

Activations Density 0.093%

superlatives, such as "most", in a given context

superlative terms indicating the highest degree or intensity of a characteristic

No Comments

No Known Activations