INDEX

Explanations

acronyms or abbreviations that end with 'WM'

oai_token-act-pair · gpt-3.5-turbo

abbreviations or acronyms related to technical specifications and components

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

ways

-0.65

schild

-0.63

ENE

-0.61

 compr

-0.54

 striking

-0.54

ALLY

-0.54

say

-0.53

 tsunami

-0.53

ahime

-0.53

aspx

-0.53

POSITIVE LOGITS

Cola

0.88

ucket

0.80

 IMAGES

0.79

ongyang

0.76

 nipples

0.75

atche

0.75

Bear

0.72

umatic

0.72

sylvania

0.70

Brother

0.69

Activations Density 0.102%

acronyms or abbreviations that end with 'WM'

abbreviations or acronyms related to technical specifications and components

No Comments

No Known Activations