INDEX

Explanations

phrases related to follow-up actions or events

oai_token-act-pair · gpt-3.5-turbo

references to follow-up actions or discussions

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

BIT

-0.79

renheit

-0.78

bis

-0.74

Hub

-0.71

gemony

-0.68

Football

-0.65

soDeliveryDate

-0.65

glers

-0.65

avery

-0.65

escape

-0.64

POSITIVE LOGITS

 observations

0.77

 observation

0.76

spection

0.75

dates

0.75

 surgeries

0.75

 inspection

0.74

noon

0.74

 examinations

0.74

 examination

0.73

iminary

0.73

Activations Density 0.049%

phrases related to follow-up actions or events

references to follow-up actions or discussions

No Comments

No Known Activations