INDEX

Explanations

The neuron is activating for words related to steps or instructions represented by words like "Next"

oai_token-act-pair · gpt-3.5-turbo

instances of the word "Next."

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 Feldman

-0.75

 sexes

-0.67

 brim

-0.66

 Sinai

-0.64

utic

-0.60

iveness

-0.60

 Ãĸ

-0.60

 Erie

-0.59

ted

-0.59

 heterogeneity

-0.59

POSITIVE LOGITS

ĻĤ

0.80

Next

0.78

door

0.76

Scene

0.73

 millenn

0.72

 installment

0.69

ļéĨĴ

0.69

 week

0.68

Phase

0.68

phase

0.68

Activations Density 0.035%

The neuron is activating for words related to steps or instructions represented by words like "Next"

instances of the word "Next."

No Comments

No Known Activations