INDEX

Explanations

names or terms related to language or regions, particularly including the phoneme 'ang'

oai_token-act-pair · gpt-3.5-turbo

combinations of letters or phonemes resembling abbreviations or acronyms

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

ãĤ©

-0.70

ĻĤ

-0.70

scription

-0.65

cess

-0.65

Bloomberg

-0.61

 quotas

-0.61

PDATE

-0.61

cling

-0.61

Applic

-0.61

 bene

-0.61

POSITIVE LOGITS

orno

0.78

gow

0.77

adesh

0.70

ampa

0.66

hiro

0.64

har

0.64

oland

0.63

entle

0.63

ribune

0.62

ratulations

0.62

Activations Density 0.235%

names or terms related to language or regions, particularly including the phoneme 'ang'

combinations of letters or phonemes resembling abbreviations or acronyms

No Comments

No Known Activations