INDEX

Explanations

instances where something happens repeatedly or sequentially

oai_token-act-pair · gpt-3.5-turbo

phrases indicating repetition or emphasizing singularity

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 4-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.4.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.4.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

 Lazarus

-0.66

aman

-0.65

mun

-0.62

LAB

-0.60

 Mirage

-0.59

scroll

-0.58

eret

-0.58

orter

-0.57

uty

-0.57

wm

-0.56

POSITIVE LOGITS

 anymore

0.98

nor

0.91

 mention

0.77

 slightest

0.76

 anywhere

0.75

 necessarily

0.73

 describ

0.69

 anything

0.68

 anybody

0.64

 coincidence

0.63

Activations Density 0.312%

instances where something happens repeatedly or sequentially

phrases indicating repetition or emphasizing singularity

No Comments

No Known Activations