INDEX

Explanations

phrases indicating composition or structure

oai_token-act-pair · gpt-3.5-turbo

phrases that describe components or parts of something

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

 confir

-0.80

soType

-0.78

 undermin

-0.72

 override

-0.70

 mosqu

-0.69

 showc

-0.68

itability

-0.67

visor

-0.65

orah

-0.64

aware

-0.64

POSITIVE LOGITS

 sorts

0.78

umbn

0.70

Cly

0.67

 varying

0.63

 Eucl

0.63

éĹĺ

0.62

course

0.61

++++++++

0.61

 course

0.59

 three

0.59

Activations Density 0.056%

phrases indicating composition or structure

phrases that describe components or parts of something

No Comments

No Known Activations