INDEX

Explanations

In this case, the neuron seems to be identifying specific characters or patterns, possibly related to a different language or encoding, as there are non-standard characters present in the text

oai_token-act-pair · gpt-3.5-turbo

symbols or non-standard characters with specific high frequencies

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 9-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.9.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.9.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

 loopholes

-0.75

 chunks

-0.71

 chess

-0.69

 unfocused

-0.68

 spoiler

-0.68

 mainline

-0.68

 multiplication

-0.68

 disruption

-0.68

 unborn

-0.68

 distraction

-0.67

POSITIVE LOGITS

ï¸ı

1.28

ng

1.11

sa

1.11

ti

1.10

1.09

1.06

ÃŃ

1.04

Å«

1.03

tal

1.03

Ä

1.03

Activations Density 0.064%

In this case, the neuron seems to be identifying specific characters or patterns, possibly related to a different language or encoding, as there are non-standard characters present in the text

symbols or non-standard characters with specific high frequencies

No Comments

No Known Activations