INDEX

Explanations

phrases indicating a progression or transition from one state to another

oai_token-act-pair · gpt-3.5-turbo

phrases indicating progressive actions or processes

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

ids

-0.72

juven

-0.69

 immortal

-0.65

pur

-0.65

 blot

-0.64

rug

-0.63

 sort

-0.62

 stand

-0.62

 quake

-0.61

 mistaken

-0.59

POSITIVE LOGITS

Through

1.02

 Through

0.99

edIn

0.81

through

0.81

ategory

0.77

clair

0.77

 thru

0.74

 Collider

0.72

cape

0.72

ensibly

0.72

Activations Density 0.007%

phrases indicating a progression or transition from one state to another

phrases indicating progressive actions or processes

No Comments

No Known Activations