INDEX

Explanations

instances of grammatical errors

oai_token-act-pair · gpt-3.5-turbo

references to grammar and linguistics

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.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

isol

-0.75

 Protective

-0.75

 Rockefeller

-0.73

oppable

-0.72

 Ster

-0.72

vr

-0.72

 Cerberus

-0.71

products

-0.71

Temperature

-0.70

arger

-0.69

POSITIVE LOGITS

 grammar

3.35

 lingu

2.40

 linguistic

2.28

 pronunciation

2.13

 Language

2.06

 phon

1.99

 spelling

1.96

 bilingual

1.88

 dialect

1.85

 parser

1.83

Activations Density 0.062%

instances of grammatical errors

references to grammar and linguistics

No Comments

No Known Activations