INDEX

Explanations

quantitative measurements appearing as fractions or percentages

oai_token-act-pair · gpt-3.5-turbo

phrases indicating proportions or ratios

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.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

Few

-0.66

andr

-0.66

few

-0.63

 Cosponsors

-0.62

Middle

-0.62

too

-0.61

-0.60

among

-0.60

-0.59

strong

-0.59

POSITIVE LOGITS

GDP

1.05

 total

0.91

 what

0.84

 capacity

0.79

its

0.74

 normal

0.71

 maximum

0.71

 original

0.68

xon

0.68

 their

0.65

Activations Density 0.095%

quantitative measurements appearing as fractions or percentages

phrases indicating proportions or ratios

No Comments

No Known Activations