INDEX

Explanations

specific strings of text with a combination of letters and numbers

oai_token-act-pair · gpt-3.5-turbo

abbreviations or acronyms related to media outlets

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

ãĥ¼ãĥĨ

-0.87

querque

-0.87

 Marketable

-0.79

ħĭ

-0.77

INAL

-0.75

 guiActiveUn

-0.69

Pear

-0.68

ãĥł

-0.66

rities

-0.66

 srfAttach

-0.65

POSITIVE LOGITS

nesday

1.10

ÅĤ

0.78

itzer

0.70

 Barrett

0.70

bledon

0.68

wright

0.68

hof

0.67

nant

0.66

 Norton

0.66

boys

0.66

Activations Density 0.141%

specific strings of text with a combination of letters and numbers

abbreviations or acronyms related to media outlets

No Comments

No Known Activations