INDEX

Explanations

numeric values or characters in a specific format

oai_token-act-pair · gpt-3.5-turbo

inquiries or references to singular and plural nouns, particularly in questions or discussions

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

 helicop

-0.71

emale

-0.69

Keefe

-0.61

 Gibraltar

-0.59

 quota

-0.58

 surn

-0.58

anamo

-0.58

 Borders

-0.57

 Seym

-0.57

 Azerb

-0.57

POSITIVE LOGITS

STAR

0.66

 Entered

0.65

 SHARES

0.64

 largeDownload

0.61

 TEXT

0.61

 âĢº

0.59

 FILE

0.58

CRIPTION

0.57

 SUPPORT

0.57

 ROCK

0.57

Activations Density 0.181%

numeric values or characters in a specific format

inquiries or references to singular and plural nouns, particularly in questions or discussions

No Comments

No Known Activations