INDEX

Explanations

The main thing this neuron does is find words related to options or choices

oai_token-act-pair · gpt-3.5-turbo

phrases that express alternatives or conditionality

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

enger

-0.63

avan

-0.61

 Peak

-0.58

imen

-0.57

oro

-0.57

oya

-0.56

auri

-0.55

eva

-0.55

risis

-0.54

erb

-0.54

POSITIVE LOGITS

 preferably

0.81

 evidenced

0.65

etheless

0.65

 congr

0.63

aukee

0.61

artifacts

0.60

especially

0.59

cause

0.56

 maybe

0.56

 EVEN

0.56

Activations Density 0.548%

The main thing this neuron does is find words related to options or choices

phrases that express alternatives or conditionality

No Comments

No Known Activations