INDEX

Explanations

words related to physical features and attributes

oai_token-act-pair · gpt-3.5-turbo

concepts related to organization and structure

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 7-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.7.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.7.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

cale

-0.73

ynski

-0.72

uden

-0.72

 Tribunal

-0.64

nesday

-0.62

mma

-0.61

scl

-0.59

Il

-0.58

aurus

-0.58

 Gazette

-0.56

POSITIVE LOGITS

busters

0.94

lessly

0.94

wise

0.88

less

0.86

breakers

0.84

able

0.75

ishly

0.75

ably

0.74

iques

0.74

ically

0.73

Activations Density 0.705%

words related to physical features and attributes

concepts related to organization and structure

No Comments

No Known Activations