INDEX

Explanations

technical numerical data and references

oai_token-act-pair · gpt-3.5-turbo

numerical references, particularly percentages and years

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 7-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.7.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.7.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

gets

-0.84

maid

-0.70

isters

-0.69

grass

-0.68

unks

-0.67

wich

-0.67

meet

-0.67

arent

-0.66

ener

-0.65

uden

-0.64

POSITIVE LOGITS

Ñĭ

0.93

ropolitan

0.78

0.77

âĸĪ

0.74

%:

0.72

eers

0.71

chan

0.71

rpm

0.69

eenth

0.69

0.68

Activations Density 0.065%

technical numerical data and references

numerical references, particularly percentages and years

No Comments

No Known Activations