Skip to content

kev-finetune: list Bifrost among data-generation endpoints #272

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
Swpn0neel wants to merge 1 commit into
base: main
Choose a base branch
from
+4 −4
Open

kev-finetune: list Bifrost among data-generation endpoints #272

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
6 changes: 3 additions & 3 deletions skills/kev-finetune/SKILL.md
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -48,9 +48,9 @@ Do not generate anything before you can answer these. Ask what you cannot infer;
Even 50 real rows matter: they become the development set or the style reference for generation. If yes, plan to
use `convert_data.py`.
4. **Data source for the rest.** Offer, in this order: (a) convert existing labels; (b) an LLM writes records from the
spec (`generate_data.py`; ask which endpoint and key they want to use: OpenAI, Vercel AI Gateway, Ollama, or any
OpenAI-compatible URL); (c) you write records yourself from the `--dry-run` prompt in batches of 20 (fine for the
first 100, slow beyond).
spec (`generate_data.py`; ask which endpoint and key they want to use: OpenAI, Vercel AI Gateway,
[Bifrost](https://github.com/maximhq/bifrost), Ollama, or any OpenAI-compatible URL); (c) you write records yourself
from the `--dry-run` prompt in batches of 20 (fine for the first 100, slow beyond).
5. **Base size.** Recommend `jaredpalmer/kev-4b` (~12 min and ~$1 per run for 400 records, ~15 min for 1000).
`kev-0.8b` for fast loops, `kev-9b` for the final model. Same recipe on all three; a run transfers unchanged.
6. **Deployment and money.** Modal account ready (`modal setup`)? Budget: each `train` prints a cost bound before it
Expand Down
2 changes: 1 addition & 1 deletion skills/kev-finetune/references/data-generation.md
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -33,7 +33,7 @@ Then decide what the real rows are for:

```bash
export KEV_GEN_API_KEY=... # or OPENAI_API_KEY; AI_GATEWAY_API_KEY switches the base URL to the gateway automatically
export KEV_GEN_BASE_URL=https://api.openai.com/v1 # https://ai-gateway.vercel.sh/v1, http://localhost:11434/v1 (Ollama), any OpenAI-compatible URL
export KEV_GEN_BASE_URL=https://api.openai.com/v1 # https://ai-gateway.vercel.sh/v1, http://localhost:8080/v1 (Bifrost), http://localhost:11434/v1 (Ollama), any OpenAI-compatible URL
python3 scripts/generate_data.py workload.json --n 1000 --out data/x.jsonl --model gpt-4.1-mini --examples data/x.real.jsonl
```

Expand Down