INDEX

Explanations

capital letters followed by numbers, possibly representing specific codes or identifiers

oai_token-act-pair · gpt-3.5-turbo

letters that are likely part of 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 @ 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

 Leilan

-0.67

ioxide

-0.63

 explanations

-0.62

 grounds

-0.62

 steps

-0.61

 wrongful

-0.59

 pages

-0.59

 pockets

-0.59

furt

-0.59

Zin

-0.59

POSITIVE LOGITS

ACTED

0.95

BR

0.85

cellence

0.84

Ì¶

0.84

OD

0.78

VP

0.78

OT

0.75

\)

0.75

BS

0.75

DS

0.75

Activations Density 0.140%

capital letters followed by numbers, possibly representing specific codes or identifiers

letters that are likely part of abbreviations or acronyms

No Comments

No Known Activations