INDEX

Explanations

It seems that Neuron 4 may be looking for steps or instructions in a process

oai_token-act-pair · gpt-3.5-turbo

procedural steps in instructional content

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

inately

-0.82

ciating

-0.79

rums

-0.74

iqueness

-0.72

ores

-0.70

 Pengu

-0.69

 flourishing

-0.68

 constit

-0.68

ãĥīãĥ©ãĤ´ãĥ³

-0.68

 subsidized

-0.65

POSITIVE LOGITS

Step

1.05

 Steps

1.00

hani

0.94

Flo

0.90

daughter

0.88

hens

0.87

sis

0.81

dad

0.80

iblings

0.78

 Step

0.78

Activations Density 0.017%

It seems that Neuron 4 may be looking for steps or instructions in a process

procedural steps in instructional content

No Comments

No Known Activations