INDEX

Explanations

numbers and comparisons

oai_token-act-pair · gpt-3.5-turbo

the word "slightly" and its variations, indicating nuances or moderate changes

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

iens

-0.78

 Aviv

-0.76

emy

-0.74

velt

-0.72

elsen

-0.70

front

-0.68

yers

-0.67

 Feast

-0.67

uments

-0.65

¥µ

-0.65

POSITIVE LOGITS

 offset

0.81

 overlap

0.79

 tint

0.73

 tang

0.72

0.69

 inaccurate

0.69

 incl

0.68

otropic

0.68

 insignificant

0.68

aditional

0.67

Activations Density 0.012%

numbers and comparisons

the word "slightly" and its variations, indicating nuances or moderate changes

No Comments

No Known Activations