INDEX

Explanations

questions or requests for information

oai_token-act-pair · gpt-3.5-turbo

commands or phrases that inquire for information or prompt questions

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

cutting

-0.72

 Tigers

-0.72

Ĥ¬

-0.68

pite

-0.64

 Formation

-0.64

swing

-0.63

rient

-0.62

ffen

-0.62

clude

-0.61

abama

-0.61

POSITIVE LOGITS

naires

1.04

 rhet

1.04

 probing

0.98

 questions

0.97

wered

0.88

 asked

0.84

ask

0.80

 answered

0.79

 asking

0.79

 politely

0.79

Activations Density 0.043%

questions or requests for information

commands or phrases that inquire for information or prompt questions

No Comments

No Known Activations