INDEX

Explanations

numbers and symbols in a specific format

oai_token-act-pair · gpt-3.5-turbo

numeric identifiers or codes related to games or scores

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.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

 bonded

-0.69

 deadlines

-0.69

ailability

-0.65

 actresses

-0.65

vow

-0.62

 radio

-0.61

psons

-0.61

 palate

-0.60

Ĥ¬

-0.59

agna

-0.58

POSITIVE LOGITS

 Tactics

0.89

 Mortal

0.83

0.80

+.

0.77

fb

0.76

+,

0.76

ace

0.74

nian

0.72

âĢ³

0.72

rn

0.72

Activations Density 0.047%

numbers and symbols in a specific format

numeric identifiers or codes related to games or scores

No Comments

No Known Activations