INDEX

Explanations

words related to guiding instructions or advice

oai_token-act-pair · gpt-3.5-turbo

phrases that initiate discussions or provide context about conditions or examples

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

 swat

-0.52

 caps

-0.50

 ï¿½

-0.48

 drib

-0.47

 opposite

-0.47

 ..."

-0.45

 diam

-0.45

 scouting

-0.45

 crystal

-0.44

``

-0.44

POSITIVE LOGITS

anmar

0.77

agascar

0.76

resa

0.76

pherd

0.75

raltar

0.70

bidden

0.69

ngth

0.67

odore

0.67

inburgh

0.65

icularly

0.63

Activations Density 0.739%

words related to guiding instructions or advice

phrases that initiate discussions or provide context about conditions or examples

No Comments

No Known Activations