INDEX

Explanations

numerical time durations mentioned in years

oai_token-act-pair · gpt-3.5-turbo

references to durations of time, particularly in years

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

romeda

-0.77

erno

-0.76

 hypocr

-0.72

Osw

-0.71

eah

-0.67

irie

-0.66

ãĥ´ãĤ¡

-0.66

liction

-0.65

anooga

-0.64

este

-0.63

POSITIVE LOGITS

ago

1.24

Ago

0.99

 apiece

0.97

 olds

0.90

 increments

0.84

 onwards

0.83

 elapsed

0.83

 onward

0.80

duration

0.73

/$

0.72

Activations Density 0.142%

numerical time durations mentioned in years

references to durations of time, particularly in years

No Comments

No Known Activations