INDEX

Explanations

code-related strings or snippets

oai_token-act-pair · gpt-3.5-turbo

numerical representations and sequences

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.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

 condol

-0.77

chwitz

-0.71

 Lauder

-0.69

 Perkins

-0.65

Compat

-0.64

 Coulter

-0.63

yre

-0.63

 distressed

-0.62

 cens

-0.61

 Wiley

-0.61

POSITIVE LOGITS

=-=-=-=-

0.77

 <!--

0.71

izo

0.67

itate

0.65

 Sharks

0.62

odo

0.61

;;

0.61

ulas

0.60

Air

0.60

izza

0.60

Activations Density 0.041%

code-related strings or snippets

numerical representations and sequences

No Comments

No Known Activations