INDEX

Explanations

words related to advice or instructions

oai_token-act-pair · gpt-3.5-turbo Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res_fs24576-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_24576

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

Built

-0.81

Specific

-0.79

ãĥ¼ãĥĨãĤ£

-0.73

NetMessage

-0.71

Frag

-0.70

Reading

-0.66

Title

-0.65

Primary

-0.65

Rap

-0.65

MSN

-0.65

POSITIVE LOGITS

 syndrome

0.95

!,

0.86

âĦ¢

0.83

 fallacy

0.81

 Productions

0.73

 Syndrome

0.71

!.

0.70

ppa

0.70

 mentality

0.70

0.69

Activations Density 0.624%

words related to advice or instructions

No Comments

No Known Activations