INDEX

Explanations

mentions of escaping or escape-related actions

oai_token-act-pair · gpt-3.5-turbo

instances of the word "escape" and its variations, particularly in the context of fleeing or being freed from a situation

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

 stead

-0.63

ificent

-0.62

inki

-0.61

spir

-0.61

ftime

-0.60

papers

-0.60

¾

-0.59

 ledger

-0.59

abouts

-0.58

iann

-0.58

POSITIVE LOGITS

 detection

0.98

 unsc

0.93

 captivity

0.90

 confinement

0.86

 from

0.85

 capture

0.84

 punishment

0.84

 into

0.81

 prosecution

0.78

 justice

0.74

Activations Density 0.054%

mentions of escaping or escape-related actions

instances of the word "escape" and its variations, particularly in the context of fleeing or being freed from a situation

No Comments

No Known Activations