f2f0e3b
f2f0e3b42bfd3f149198979dfa049e457e8a675a3a411ea
c7f2be4| Status | File | + | - |
|---|---|---|---|
| A |
.vscode/launch.json
|
15 | 0 |
| A |
.vscode/tasks.json
|
11 | 0 |
| A |
test/fixtures/sample.dlm
|
38 | 0 |
.vscode/launch.jsonadded@@ -0,0 +1,15 @@ | ||
| 1 | +{ | |
| 2 | + "version": "0.2.0", | |
| 3 | + "configurations": [ | |
| 4 | + { | |
| 5 | + "name": "Run Extension", | |
| 6 | + "type": "extensionHost", | |
| 7 | + "request": "launch", | |
| 8 | + "args": [ | |
| 9 | + "--extensionDevelopmentPath=${workspaceFolder}" | |
| 10 | + ], | |
| 11 | + "outFiles": ["${workspaceFolder}/out/**/*.js"], | |
| 12 | + "preLaunchTask": "npm: build" | |
| 13 | + } | |
| 14 | + ] | |
| 15 | +} | |
.vscode/tasks.jsonadded@@ -0,0 +1,11 @@ | ||
| 1 | +{ | |
| 2 | + "version": "2.0.0", | |
| 3 | + "tasks": [ | |
| 4 | + { | |
| 5 | + "type": "npm", | |
| 6 | + "script": "build", | |
| 7 | + "group": "build", | |
| 8 | + "problemMatcher": [] | |
| 9 | + } | |
| 10 | + ] | |
| 11 | +} | |
test/fixtures/sample.dlmadded@@ -0,0 +1,38 @@ | ||
| 1 | +--- | |
| 2 | +dlm_id: 01KPQ9M3000000000000000000 | |
| 3 | +dlm_version: 15 | |
| 4 | +base_model: smollm2-135m | |
| 5 | +training: | |
| 6 | + adapter: lora | |
| 7 | + lora_r: 16 | |
| 8 | + learning_rate: 2e-4 | |
| 9 | + num_epochs: 3 | |
| 10 | +--- | |
| 11 | +# Sample Document | |
| 12 | + | |
| 13 | +This is a sample DLM document for testing the VSCode extension. | |
| 14 | + | |
| 15 | +::instruction:: | |
| 16 | +### Q | |
| 17 | +What is a Document Language Model? | |
| 18 | + | |
| 19 | +### A | |
| 20 | +A .dlm file is a single UTF-8 text file that becomes a local, reproducible, | |
| 21 | +trainable LLM. Edit the document, retrain, share. | |
| 22 | + | |
| 23 | +::instruction:: | |
| 24 | +### Q | |
| 25 | +How do you train a DLM? | |
| 26 | + | |
| 27 | +### A | |
| 28 | +Run `dlm train your-file.dlm` and the adapter trains on the document content. | |
| 29 | + | |
| 30 | +::preference:: | |
| 31 | +### Prompt | |
| 32 | +Explain LoRA in one sentence. | |
| 33 | + | |
| 34 | +### Chosen | |
| 35 | +LoRA adds small trainable matrices to frozen model layers, enabling efficient fine-tuning. | |
| 36 | + | |
| 37 | +### Rejected | |
| 38 | +LoRA is a method for training language models that involves modifying the architecture of the model by introducing additional parameters in the form of low-rank decomposition matrices that are applied to the attention weight matrices, which allows for parameter-efficient fine-tuning while keeping the original pre-trained weights frozen. | |