Skip to content

Commit 4db2a9b

Browse files
Add walkthroughs and delete tutorial (substratusai#20)
* move content from tutorial to walkthrough * update introduction --------- Co-authored-by: Brandon J. Bjelland <[email protected]>
1 parent 41c01e7 commit 4db2a9b

17 files changed

+1061
-1160
lines changed

.vscode/settings.json

Lines changed: 5 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,5 @@
1+
{
2+
"cSpell.words": [
3+
"finetune"
4+
]
5+
}

docs/introduction.md

Lines changed: 5 additions & 3 deletions
Original file line numberDiff line numberDiff line change
@@ -9,11 +9,13 @@ Substratus is a cross-cloud substrate for training and serving ML models. Substr
99

1010
## Why Substratus?
1111

12-
* Train and serve models from within your cloud account, on a portable platform. Your data stays private.
12+
* Zero code needed to deploy and finetune OSS LLMs
13+
* Ability to deploy and finetune state of the art large language models easily within an hour.
14+
* Automatically provision all the required infrastructure (K8s cluster, buckets etc) in your cloud account using Terraform.
15+
* Scale from 0 and spot VMs supported by default (GPUs $$$).
16+
* Your data stays private and won't be used by the corporate overlords
1317
* Leverage containers to avoid library lock-in and dependency wrangling.
14-
* Let substratus calculate your resource requirements and automatically provision GPUs, CPUs, Storage, and Memory.
1518
* Adopt best practice conventions by default.
16-
* Train pre-packaged state of the art models on your own datasets.
1719
* Leverage GitOps out of the box.
1820

1921
## One Minute Intro Video

docs/quickstart.ipynb

Lines changed: 33 additions & 42 deletions
Original file line numberDiff line numberDiff line change
@@ -229,54 +229,46 @@
229229
},
230230
{
231231
"cell_type": "code",
232-
"execution_count": 10,
232+
"execution_count": null,
233233
"id": "90825af1",
234234
"metadata": {},
235-
"outputs": [
236-
{
237-
"name": "stdout",
238-
"output_type": "stream",
239-
"text": [
240-
" % Total % Received % Xferd Average Speed Time Time Time Current\n",
241-
" Dload Upload Total Spent Left Speed\n",
242-
" 0 0 0 0 0 0 0 0 --:--:-- --:--:-- --:--:-- 0"
243-
]
244-
},
245-
{
246-
"name": "stdout",
247-
"output_type": "stream",
248-
"text": [
249-
"100 458 100 326 100 132 171 69 0:00:01 0:00:01 --:--:-- 241\n",
250-
"\u001b[1;39m{\n",
251-
" \u001b[0m\u001b[34;1m\"id\"\u001b[0m\u001b[1;39m: \u001b[0m\u001b[0;32m\"cmpl-e42772faf58cd46c18a955f1\"\u001b[0m\u001b[1;39m,\n",
252-
" \u001b[0m\u001b[34;1m\"object\"\u001b[0m\u001b[1;39m: \u001b[0m\u001b[0;32m\"text_completion\"\u001b[0m\u001b[1;39m,\n",
253-
" \u001b[0m\u001b[34;1m\"created\"\u001b[0m\u001b[1;39m: \u001b[0m\u001b[0;39m1689485483\u001b[0m\u001b[1;39m,\n",
254-
" \u001b[0m\u001b[34;1m\"model\"\u001b[0m\u001b[1;39m: \u001b[0m\u001b[0;32m\"falcon-7b-instruct\"\u001b[0m\u001b[1;39m,\n",
255-
" \u001b[0m\u001b[34;1m\"choices\"\u001b[0m\u001b[1;39m: \u001b[0m\u001b[1;39m[\n",
256-
" \u001b[1;39m{\n",
257-
" \u001b[0m\u001b[34;1m\"text\"\u001b[0m\u001b[1;39m: \u001b[0m\u001b[0;32m\"\\nGeorge Washington was the first president of the United States.\"\u001b[0m\u001b[1;39m,\n",
258-
" \u001b[0m\u001b[34;1m\"index\"\u001b[0m\u001b[1;39m: \u001b[0m\u001b[0;39m0\u001b[0m\u001b[1;39m,\n",
259-
" \u001b[0m\u001b[34;1m\"logprobs\"\u001b[0m\u001b[1;39m: \u001b[0m\u001b[1;30mnull\u001b[0m\u001b[1;39m,\n",
260-
" \u001b[0m\u001b[34;1m\"finish_reason\"\u001b[0m\u001b[1;39m: \u001b[0m\u001b[0;32m\"length\"\u001b[0m\u001b[1;39m\n",
261-
" \u001b[1;39m}\u001b[0m\u001b[1;39m\n",
262-
" \u001b[1;39m]\u001b[0m\u001b[1;39m,\n",
263-
" \u001b[0m\u001b[34;1m\"usage\"\u001b[0m\u001b[1;39m: \u001b[0m\u001b[1;39m{\n",
264-
" \u001b[0m\u001b[34;1m\"prompt_tokens\"\u001b[0m\u001b[1;39m: \u001b[0m\u001b[0;39m11\u001b[0m\u001b[1;39m,\n",
265-
" \u001b[0m\u001b[34;1m\"completion_tokens\"\u001b[0m\u001b[1;39m: \u001b[0m\u001b[0;39m12\u001b[0m\u001b[1;39m,\n",
266-
" \u001b[0m\u001b[34;1m\"total_tokens\"\u001b[0m\u001b[1;39m: \u001b[0m\u001b[0;39m23\u001b[0m\u001b[1;39m\n",
267-
" \u001b[1;39m}\u001b[0m\u001b[1;39m\n",
268-
"\u001b[1;39m}\u001b[0m\n"
269-
]
270-
}
271-
],
235+
"outputs": [],
272236
"source": [
273237
"! curl http://localhost:8080/v1/completions \\\n",
274238
" -H \"Content-Type: application/json\" \\\n",
275239
" -d '{ \\\n",
276240
" \"model\": \"falcon-7b-instruct\", \\\n",
277241
" \"prompt\": \"Who was the first president of the United States? \", \\\n",
278-
" \"max_tokens\": 12\\\n",
279-
" }' | jq"
242+
" \"max_tokens\": 10\\\n",
243+
" }' "
244+
]
245+
},
246+
{
247+
"cell_type": "markdown",
248+
"id": "59cdbd6c",
249+
"metadata": {},
250+
"source": [
251+
"```json\n",
252+
"{\n",
253+
" \"id\": \"cmpl-e42772faf58cd46c18a955f1\",\n",
254+
" \"object\": \"text_completion\",\n",
255+
" \"created\": 1689485483,\n",
256+
" \"model\": \"falcon-7b-instruct\",\n",
257+
" \"choices\": [\n",
258+
" {\n",
259+
" \"text\": \"\\nGeorge Washington was the first president of the United States.\",\n",
260+
" \"index\": 0,\n",
261+
" \"logprobs\": null,\n",
262+
" \"finish_reason\": \"length\"\n",
263+
" }\n",
264+
" ],\n",
265+
" \"usage\": {\n",
266+
" \"prompt_tokens\": 11,\n",
267+
" \"completion_tokens\": 12,\n",
268+
" \"total_tokens\": 23\n",
269+
" }\n",
270+
"}\n",
271+
"```"
280272
]
281273
},
282274
{
@@ -285,8 +277,7 @@
285277
"id": "95a9966d-5679-4208-9b28-0323aa80cd79",
286278
"metadata": {},
287279
"source": [
288-
"If you are interested in continuing your journey through Substratus, take a look at the [Guided Walkthrough](./category/walkthrough) or\n",
289-
"follow the [tutorial to finetune falcon-7b-instruct](./tutorials/deploying-finetuning-falcon-7b-instruct.md) with a custom dataset."
280+
"If you are interested in continuing your journey through Substratus, take a look at the [Guided Walkthrough](./category/walkthrough) to learn how to finetune models with your own dataset and much more!"
290281
]
291282
},
292283
{

docs/quickstart.md

Lines changed: 25 additions & 28 deletions
Original file line numberDiff line numberDiff line change
@@ -103,36 +103,33 @@ All substratus Servers ship with an API and interactive frontend. Open up your b
103103
-d '{ \
104104
"model": "falcon-7b-instruct", \
105105
"prompt": "Who was the first president of the United States? ", \
106-
"max_tokens": 12\
107-
}' | jq
106+
"max_tokens": 10\
107+
}'
108108
```
109109

110-
% Total % Received % Xferd Average Speed Time Time Time Current
111-
Dload Upload Total Spent Left Speed
112-
0 0 0 0 0 0 0 0 --:--:-- --:--:-- --:--:-- 0100 458 100 326 100 132 171 69 0:00:01 0:00:01 --:--:-- 241
113-
{
114-
"id": "cmpl-e42772faf58cd46c18a955f1",
115-
"object": "text_completion",
116-
"created": 1689485483,
117-
"model": "falcon-7b-instruct",
118-
"choices": [
119-
{
120-
"text": "\nGeorge Washington was the first president of the United States.",
121-
"index": 0,
122-
"logprobs": null,
123-
"finish_reason": "length"
124-
}
125-
],
126-
"usage": {
127-
"prompt_tokens": 11,
128-
"completion_tokens": 12,
129-
"total_tokens": 23
130-
}
131-
}
132-
133-
134-
If you are interested in continuing your journey through Substratus, take a look at the [Guided Walkthrough](./category/walkthrough) or
135-
follow the [tutorial to finetune falcon-7b-instruct](./tutorials/deploying-finetuning-falcon-7b-instruct.md) with a custom dataset.
110+
```json
111+
{
112+
"id": "cmpl-e42772faf58cd46c18a955f1",
113+
"object": "text_completion",
114+
"created": 1689485483,
115+
"model": "falcon-7b-instruct",
116+
"choices": [
117+
{
118+
"text": "\nGeorge Washington was the first president of the United States.",
119+
"index": 0,
120+
"logprobs": null,
121+
"finish_reason": "length"
122+
}
123+
],
124+
"usage": {
125+
"prompt_tokens": 11,
126+
"completion_tokens": 12,
127+
"total_tokens": 23
128+
}
129+
}
130+
```
131+
132+
If you are interested in continuing your journey through Substratus, take a look at the [Guided Walkthrough](./category/walkthrough) to learn how to finetune models with your own dataset and much more!
136133

137134
## Cleanup
138135

docs/tutorials/_category_.json

Lines changed: 0 additions & 7 deletions
This file was deleted.

0 commit comments

Comments
 (0)