INDEX

Explanations

phrases indicating continued actions or sequences

oai_token-act-pair · gpt-3.5-turbo

phrases related to continuous actions or processes

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

 Surprise

-0.86

 Gleaming

-0.76

 Mesa

-0.73

 Yosemite

-0.72

 Shelter

-0.72

 Runner

-0.70

 Abdel

-0.70

dylib

-0.68

Runner

-0.64

 Clothing

-0.63

POSITIVE LOGITS

 uninterrupted

0.81

 adolesc

0.77

tions

0.76

clusively

0.75

abre

0.74

edly

0.73

 unab

0.69

 indefinitely

0.69

 diligence

0.68

nces

0.67

Activations Density 0.238%

phrases indicating continued actions or sequences

phrases related to continuous actions or processes

No Comments

No Known Activations