INDEX

Explanations

phrases related to causality and action

oai_token-act-pair · gpt-3.5-turbo

instances of the word "through."

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

etus

-0.69

thy

-0.64

Ont

-0.63

ETF

-0.62

ãĤ·

-0.61

intend

-0.61

CTV

-0.61

atron

-0.61

qua

-0.60

anooga

-0.59

POSITIVE LOGITS

 sheer

1.19

 intermedi

1.05

 attrition

1.04

 inaction

0.91

 proxies

0.90

OUT

0.88

 obscurity

0.87

gone

0.83

 partnerships

0.82

 clenched

0.82

Activations Density 0.060%

phrases related to causality and action

instances of the word "through."

No Comments

No Known Activations