INDEX

Explanations

programming-related terms or syntax

oai_token-act-pair · gpt-3.5-turbo

underscore-prefixed identifiers or terms related to programming, particularly in C++ or similar languages

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 4-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.4.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.4.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

 Doctors

-0.80

 Pearce

-0.74

 Garry

-0.74

quished

-0.74

swick

-0.72

 Paddock

-0.71

 Mercy

-0.71

eteria

-0.71

 Infect

-0.70

 Alicia

-0.70

POSITIVE LOGITS

dict

1.15

chance

1.01

default

1.01

state

1.01

blank

0.99

gradient

0.99

usable

0.94

impl

0.94

delay

0.93

enabled

0.92

Activations Density 0.021%

programming-related terms or syntax

underscore-prefixed identifiers or terms related to programming, particularly in C++ or similar languages

No Comments

No Known Activations