INDEX

Explanations

specific strings of characters, possibly related to coding or technical documentation

oai_token-act-pair · gpt-3.5-turbo

substrings and patterns within text, particularly related to code or technical identifiers

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.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

 ultras

-0.62

 Seat

-0.62

 Instr

-0.61

EVA

-0.60

 Immunity

-0.60

enegger

-0.60

KM

-0.58

 Gemini

-0.58

Gum

-0.57

 Stim

-0.57

POSITIVE LOGITS

isoft

1.08

yssey

0.99

mble

0.96

cean

0.95

ertain

0.87

chwitz

0.87

umen

0.86

uth

0.84

vre

0.82

odox

0.81

Activations Density 0.161%

specific strings of characters, possibly related to coding or technical documentation

substrings and patterns within text, particularly related to code or technical identifiers

No Comments

No Known Activations