INDEX

Explanations

purely procedural steps or instructions

oai_token-act-pair · gpt-3.5-turbo

cooking-related actions and instructions

oai_token-act-pair · gpt-4o-mini

Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 9-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.9.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.9.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

IFrame

Link

Negative Logits

 answers

-0.85

 categ

-0.73

 fielded

-0.68

 situational

-0.65

 royalty

-0.65

 existed

-0.65

 unanswered

-0.65

 credential

-0.64

 indebted

-0.64

 treasury

-0.64

POSITIVE LOGITS

Remove

1.16

 Pour

1.13

 Slowly

1.10

 Remove

1.05

 Then

1.02

 Drain

1.01

 Stir

1.00

 Fold

0.97

 Bake

0.97

 Divide

0.95

Activations Density 0.097%

purely procedural steps or instructions

cooking-related actions and instructions

No Known Activations