INDEX

Explanations

The neuron seems to find words or phrases related to reasons or explanations

oai_token-act-pair · gpt-3.5-turbo

the presence of the verb "is" in various contexts

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

 Highlights

-0.73

icter

-0.73

tainment

-0.71

itas

-0.69

vana

-0.64

vantage

-0.64

lez

-0.64

umbs

-0.64

 congratulations

-0.63

abouts

-0.63

POSITIVE LOGITS

 unregulated

0.86

 such

0.85

 technically

0.85

 rarely

0.85

 ostensibly

0.85

 inherently

0.84

 geographically

0.84

so

0.83

 notoriously

0.83

 already

0.83

Activations Density 0.316%

The neuron seems to find words or phrases related to reasons or explanations

the presence of the verb "is" in various contexts

No Comments

No Known Activations