INDEX

Explanations

instances of numbers and identifying characteristics

oai_token-act-pair · gpt-3.5-turbo

information about historical or factual events

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

 another

-0.69

 Another

-0.66

 elsewhere

-0.62

Another

-0.60

 Nope

-0.59

 instead

-0.59

 Otherwise

-0.56

 ..........

-0.55

 sans

-0.54

 less

-0.54

POSITIVE LOGITS

 culminating

0.61

 culminated

0.58

 influencing

0.55

rued

0.55

ULTS

0.55

sequently

0.52

 contributors

0.51

 interpre

0.51

//[

0.51

 genesis

0.50

Activations Density 0.887%

instances of numbers and identifying characteristics

information about historical or factual events

No Comments

No Known Activations