INDEX

Explanations

proper nouns such as names of researchers or organizations

oai_token-act-pair · gpt-3.5-turbo

pronouns and possessive adjectives related to individuals or groups

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 11-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.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.11.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 Tradable

-0.81

etsy

-0.76

ername

-0.75

advertising

-0.75

POST

-0.74

oke

-0.70

 âĸº

-0.69

ghazi

-0.68

OTUS

-0.67

deen

-0.67

POSITIVE LOGITS

 findings

1.50

 research

1.21

 discoveries

1.20

 conclusions

1.18

 colleague

1.18

 observations

1.16

 calculations

1.13

 thesis

1.13

 resear

1.12

 colleagues

1.11

Activations Density 0.187%

proper nouns such as names of researchers or organizations

pronouns and possessive adjectives related to individuals or groups

No Comments

No Known Activations