INDEX

Explanations

phrases indicating a statement or opinion, often in quotes

oai_token-act-pair · gpt-3.5-turbo

instances of the phrase "it was" in varying contexts

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

gress

-0.76

duc

-0.62

mobile

-0.62

 Seat

-0.60

 Laurie

-0.60

 Geneva

-0.58

Major

-0.57

minist

-0.57

AAA

-0.57

 Vers

-0.56

POSITIVE LOGITS

 resembled

0.97

 amounted

0.91

unes

0.91

zbollah

0.87

 deserved

0.82

 smelled

0.82

MpServer

0.81

OULD

0.81

 shouldn

0.80

 tasted

0.79

Activations Density 0.124%

phrases indicating a statement or opinion, often in quotes

instances of the phrase "it was" in varying contexts

No Comments

No Known Activations