INDEX

Explanations

personal pronouns followed by the verb "to be" (e.g., is, am, are)

oai_token-act-pair · gpt-3.5-turbo

first-person phrases and references to actions involving "I" and "we."

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.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

natal

-0.75

angering

-0.64

nor

-0.63

eur

-0.62

likely

-0.61

Recipe

-0.61

osis

-0.61

liest

-0.61

lihood

-0.60

ollah

-0.59

POSITIVE LOGITS

go

0.87

 goes

0.81

go

0.77

eps

0.73

iter

0.68

ppo

0.66

 belong

0.65

 again

0.65

tarians

0.62

 comes

0.61

Activations Density 0.051%

personal pronouns followed by the verb "to be" (e.g., is, am, are)

first-person phrases and references to actions involving "I" and "we."

No Comments

No Known Activations