INDEX

Explanations

terms related to restrictions or limitations

oai_token-act-pair · gpt-3.5-turbo

repeated occurrences of the prefix "und."

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 2-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.2.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.2.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

anwhile

-0.88

azines

-0.83

=-=-=-=-=-=-=-=-

-0.82

 Flavoring

-0.80

ebus

-0.78

ppo

-0.78

BOOK

-0.78

asio

-0.77

ongyang

-0.77

ovie

-0.77

POSITIVE LOGITS

iscovered

0.82

ec

0.82

ications

0.82

aunted

0.80

ifferent

0.79

icator

0.78

icated

0.77

ere

0.77

esc

0.76

ploy

0.76

Activations Density 0.006%

terms related to restrictions or limitations

repeated occurrences of the prefix "und."

No Comments

No Known Activations