Skip to content

armgit / Soup

Public Imported from GitHub
0

Fine-tune LLMs from one YAML. Layer streaming trains an 8B model on a 4 GB laptop GPU.

NameLast commit date
..
__init__.py
_eval_v0550.py
_eval_v0610.py
_eval_v0650.py
_eval_v07110.py
_webhook_cli.py
ab.py
active_sample.py
adapters.py
advise.py
agent.py
airgap.py
apple_adapter.py
attest.py
audit_log.py
autopilot.py
bench.py
bom.py
build.py
can.py
card.py
chat.py
ci.py
compile_cmd.py
compile_tools.py
completions.py
cost.py
data_canary.py
data_clean.py
data_doctor.py
data_forge.py
data_mix.py
data_score.py
data_topics.py
data.py
delinearize_llama4.py
deploy.py
diagnose.py
diff.py
distill_prompt.py
doctor.py
draft.py
drift_alarm.py
edit.py
env.py
eval.py
expect.py
export.py
fetch.py
generate.py
history.py
infer.py
ingest.py
init.py
iterative_dpo.py
license_advisor.py
llama.py
local_rl.py
lock.py
loop.py
mcp.py
merge_sharded_fsdp_weights.py
merge.py
migrate.py
monitor.py
plan.py
plugins.py
probe.py
profile.py
prune_prompt.py
push.py
quantize.py
quickstart.py
ra_dit.py
recipes.py
registry.py
reward.py
rewind.py
runs.py
serve.py
ship.py
shrink.py
spectrum.py
steer.py
sweep.py
tokenizer.py
train.py
tui.py
tunability.py
ui.py
why.py