Spaces:
Sleeping
Sleeping
File size: 39,616 Bytes
2df5c63 112679c 27cf9cd 112679c 27cf9cd 112679c 2df5c63 112679c 2df5c63 112679c 2df5c63 112679c 2df5c63 112679c 2df5c63 112679c 2df5c63 27cf9cd 2df5c63 04688c1 2df5c63 0566783 2df5c63 69f89ec cd923aa 69f89ec 2df5c63 69f89ec 2df5c63 69f89ec 2df5c63 69f89ec 2df5c63 69f89ec 2df5c63 69f89ec 2df5c63 69f89ec 2df5c63 69f89ec 2df5c63 63726b6 2df5c63 69f89ec 2df5c63 | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 672 673 674 675 676 677 678 679 680 681 682 683 684 685 686 687 688 689 690 691 692 693 694 695 696 697 698 699 700 701 702 703 704 705 706 707 708 709 710 711 712 713 714 715 716 717 718 719 720 721 722 723 724 725 726 727 728 729 730 731 732 733 734 735 736 737 738 739 740 741 742 743 744 745 746 747 748 749 750 751 752 753 754 755 756 757 758 759 760 761 762 763 764 765 766 767 768 769 770 771 772 773 774 775 776 777 778 779 780 781 782 783 784 785 786 787 788 789 790 791 792 793 794 795 796 797 798 799 800 801 802 803 804 805 806 807 808 809 810 811 812 813 814 815 816 817 818 819 820 821 822 823 824 825 826 827 828 829 830 831 832 833 834 835 836 837 838 839 840 841 842 843 844 845 846 847 848 849 850 851 852 853 854 855 856 857 858 859 860 861 862 863 864 865 866 867 868 869 870 871 872 873 874 875 876 877 878 879 880 881 882 883 884 885 886 887 888 889 890 891 892 893 894 895 896 897 898 899 900 901 902 903 904 905 906 907 908 909 910 911 912 913 914 915 916 917 918 919 920 921 922 923 924 925 926 927 928 929 930 931 932 933 934 935 936 937 938 939 940 941 942 | """
kaggle/build_notebook.py
========================
Programmatically (re)builds `train_ermap_grpo_kaggle.ipynb` from scratch.
Why a builder script?
--------------------
The hand-edited notebook drifted into a fragile state across many sessions:
mixed early-stop / fixed-budget params, stale install snippets, dead pre-flight
checks, etc. This script is the single source of truth β run it once and the
notebook is regenerated as a clean, deterministic v3 layout.
Run:
python kaggle/build_notebook.py
Output:
kaggle/train_ermap_grpo_kaggle.ipynb (overwritten)
kaggle/KAGGLE_QUICKSTART.md (overwritten)
"""
from __future__ import annotations
import json
import textwrap
from pathlib import Path
# ---------------------------------------------------------------------------
# Cell helpers
# ---------------------------------------------------------------------------
def md_cell(text: str) -> dict:
return {
"cell_type": "markdown",
"metadata": {},
"source": _split_keep_newlines(text),
}
def code_cell(text: str) -> dict:
return {
"cell_type": "code",
"execution_count": None,
"metadata": {},
"outputs": [],
"source": _split_keep_newlines(text),
}
def _split_keep_newlines(text: str) -> list[str]:
"""Notebook 'source' fields expect each line to terminate with '\n'
except the last one. Splitting like this keeps `git diff` clean when
the notebook is regenerated."""
text = textwrap.dedent(text).lstrip("\n")
if not text.endswith("\n"):
text = text + "\n"
lines = text.splitlines(keepends=True)
if lines:
# The last line should NOT have a trailing newline (Jupyter convention).
if lines[-1].endswith("\n"):
lines[-1] = lines[-1].rstrip("\n")
return lines
# ---------------------------------------------------------------------------
# Cell sources
# ---------------------------------------------------------------------------
CELL_01_TITLE = """\
# ER-MAP β Doctor Agent GRPO Training (Kaggle Free-Tier Β· v3 stable)
Trains the **Doctor LLM** (Llama-3.1-8B-Instruct, 4-bit + LoRA r=16) via GRPO
with a 3-phase curriculum on Kaggle's free GPU. Designed to survive Kaggle's
pre-baked image quirks (numpy / Pillow ABI mismatches, torch + torchvision
CUDA-major mismatches, transient `unsloth_zoo` upgrades).
## TL;DR β How to run this notebook
1. **Notebook settings (right sidebar):**
- Accelerator: **GPU T4 Γ2** (or P100)
- Internet: **On**
- Persistence: Files only
2. **Kaggle Secrets** (Add-ons β Secrets):
- **Required:** `GROQ_NURSE_API_KEY`, `GROQ_PATIENT_API_KEY`,
`GROQ_EMPATHY_JUDGE_API_KEY`, `GROQ_MEDICAL_JUDGE_API_KEY`, `HF_TOKEN`
- **Optional:** `WANDB_API_KEY`
3. **Run cells 2 β 3 (sanity + REPAIR).** When cell 3 prints
`RESTART REQUIRED`, click **Run β Restart kernel**, then resume from cell 5.
4. **Run cells 5 β 11 (verify + configure + dry-run + pre-flight).** Each cell
should print an `OK` line before moving on.
5. **Run cell 13 (the long training cell, 4β6 hours).**
6. **Run cells 14 β 17 (final push + plots + inference smoke-test).**
## Curriculum + reward thresholds (this run)
Constant per-phase rolling-avg-reward bars; sustained for **3 consecutive
GRPO groups** triggers either a phase promotion or end-of-training.
| Phase | Reward target (sustained Γ3 groups) | Action when met |
|---|---|---|
| 1 β Tool Mastery | `+1.2` | force-promote to Phase 2 |
| 2 β Clinical Reasoning | `+1.1` | force-promote to Phase 3 |
| 3 β Empathetic Negotiation | `+1.0` | END TRAINING |
Why these numbers? The un-trained 8B Doctor's baseline on the same env is
`P1=+0.76, P2=+0.59, P3=+0.39`. Targets of `+1.2 / +1.1 / +1.0` correspond
to roughly `1.6Γ / 1.9Γ / 2.6Γ` improvement over baseline β a meaningful
signal but reachable inside Kaggle's 12 h session limit.
"""
CELL_02_SANITY = """\
# === CELL 2 β Sanity check (GPU + disk + python + internet) ===
# Run this FIRST. If any check fails, fix it before running the REPAIR cell.
import os, shutil, subprocess, sys, socket
print("--- GPU ---")
try:
print(subprocess.check_output(
["nvidia-smi", "--query-gpu=name,memory.total,memory.free", "--format=csv"],
timeout=10,
).decode())
except Exception as e:
print(f"nvidia-smi failed: {e}")
print("-> Set Accelerator to 'GPU T4 x2' in the right sidebar.")
print("--- Disk (/kaggle/working) ---")
total, used, free = shutil.disk_usage("/kaggle/working")
print(f" total={total/1e9:5.1f} GB | used={used/1e9:5.1f} GB | free={free/1e9:5.1f} GB")
if free < 8 * 1e9:
print(" WARNING: free disk < 8 GB β repair cell may fail. "
"Consider 'Run > Restart and clear cell outputs' to reset /tmp.")
print("--- Python ---")
print(f" python={sys.version.split()[0]} | exe={sys.executable}")
print("--- Internet (api.groq.com:443) ---")
try:
socket.create_connection(("api.groq.com", 443), timeout=5).close()
print(" reachable")
except Exception as e:
print(f" UNREACHABLE: {e}")
print(" -> Settings (right sidebar) -> Internet -> ON")
"""
CELL_03_REPAIR = """\
# === CELL 3 β REPAIR CELL (idempotent full environment rebuild) ===
# Single source of truth for ER-MAP's GPU stack. Safe to re-run. After it
# finishes you'll see one of two final lines:
#
# RESTART REQUIRED -> Run -> Restart kernel, then resume from cell 5
# REPAIR OK -> proceed directly to cell 5
#
# Note: this cell only runs shell commands and one isolated subprocess.
# It deliberately does NOT `import torch / numpy / Pillow / unsloth` in the
# kernel, so re-running it after a botched install does not poison further
# attempts.
print("=" * 72); print(" CELL 3 β REPAIR"); print("=" * 72)
# 1. Clean caches (Kaggle's /kaggle/working is only 20 GB β installs
# routinely fill it after a few re-runs).
print("[1/6] Cleaning pip + tmp + HF dataset caches...")
get_ipython().system('pip cache purge -q || true')
get_ipython().system('rm -rf /tmp/* /root/.cache/pip /root/.cache/huggingface/datasets 2>/dev/null || true')
# 2. Pin torch + torchvision to the cu128 wheel (matches Kaggle's CUDA 12.8
# base image). DON'T let pip pull a generic CUDA-13 build β that breaks
# bitsandbytes (libnvJitLink.so.13 missing) and torchvision (CUDA-major
# mismatch RuntimeError at import time).
print("[2/6] Installing torch==2.10.0 + torchvision==0.25.0 (cu128)...")
get_ipython().system('pip install -q --no-cache-dir --force-reinstall '
'torch==2.10.0 torchvision==0.25.0 '
'--index-url https://download.pytorch.org/whl/cu128')
# Write a pip constraints file so subsequent installs (bnb, unsloth, trl, etc.)
# can NEVER pull a different torch from default PyPI. Without this, step 3's
# `--force-reinstall bitsandbytes` and step 4's `unsloth` upgrade re-resolve
# torch from PyPI (currently 2.11.0), which breaks the cu128 torchvision pair.
#
# Also pin numpy to whatever Kaggle's kernel already has loaded β Kaggle's
# image puts numpy at /usr/lib/python3/dist-packages while pip writes to
# /usr/local/lib/python3.12/dist-packages, so any version drift between the
# two paths trips unsloth_zoo's strict loaded-vs-installed check at import.
import subprocess as _sp
_kernel_numpy = _sp.check_output(
[sys.executable, "-c", "import numpy; print(numpy.__version__)"],
text=True,
).strip()
print(f" detected kernel numpy = {_kernel_numpy} (will pin)")
with open("/tmp/ermap_constraints.txt", "w") as _cf:
_cf.write(f"torch==2.10.0\\ntorchvision==0.25.0\\nnumpy=={_kernel_numpy}\\n")
# 3. Reinstall bitsandbytes against the now-pinned torch.
# --no-deps because bnb just needs torch at RUNTIME (it dlopens torch's
# C++ lib) β its install-time deps don't include torch.
print("[3/6] Reinstalling bitsandbytes (--no-deps to preserve torch)...")
get_ipython().system('pip install -q --no-cache-dir --force-reinstall --no-deps bitsandbytes')
# 4. Upgrade unsloth + unsloth_zoo + trl in lockstep. unsloth and
# unsloth_zoo are released as a matched pair; if pip pulls a fresh
# unsloth_zoo against an old unsloth you get
# ImportError: cannot import name 'create_gradient_checkpointing_buffer'
# The constraint file blocks them from moving torch.
print("[4/6] Upgrading unsloth + unsloth_zoo + trl (constrained)...")
get_ipython().system('pip install -q --upgrade --no-cache-dir '
'-c /tmp/ermap_constraints.txt '
'unsloth unsloth_zoo "trl>=0.18.2"')
# 5. ER-MAP runtime deps that aren't pre-installed on Kaggle.
print("[5/6] Installing ER-MAP runtime deps (constrained)...")
get_ipython().system('pip install -q --no-cache-dir '
'-c /tmp/ermap_constraints.txt '
'"groq>=0.18.0" "huggingface_hub>=0.25.0" '
'"gymnasium>=0.29.0" "openenv-core>=0.1.0"')
# 5b. Realign the pip-managed numpy with whatever the Kaggle kernel actually
# has loaded. This force-rewrites /usr/local/lib/.../numpy at the exact
# version reported by the running interpreter, so importlib.metadata
# and `numpy.__version__` agree even if Kaggle ships its base numpy at
# a different dist-packages path.
print(f"[5b/6] Realigning pip-managed numpy to {_kernel_numpy}...")
get_ipython().system(
f'pip install -q --force-reinstall --no-deps "numpy=={_kernel_numpy}"'
)
# 6. Verify in a SUBPROCESS (so the parent kernel never imports any of these
# while pip is mid-flight, which is what causes the
# 'numpy was upgraded mid-session (loaded: X, installed: Y)' RuntimeError
# we kept hitting before).
print("[6/6] Verifying via subprocess...")
import subprocess, sys, json
verify_script = r'''
import json, sys
out = {"ok": True, "details": {}, "errors": []}
try:
import importlib.metadata as md
for pkg in ("torch", "torchvision", "bitsandbytes", "unsloth", "unsloth_zoo",
"trl", "transformers", "peft", "accelerate", "groq",
"huggingface_hub", "gymnasium", "numpy", "Pillow"):
try:
out["details"][pkg + "_installed"] = md.version(pkg)
except md.PackageNotFoundError:
out["details"][pkg + "_installed"] = None
import torch, torchvision, numpy as np, PIL, unsloth, unsloth_zoo, bitsandbytes, trl
out["details"]["torch_loaded"] = torch.__version__
out["details"]["torch_cuda"] = torch.version.cuda
out["details"]["cuda_available"] = bool(torch.cuda.is_available())
out["details"]["gpu_count"] = int(torch.cuda.device_count())
out["details"]["torchvision_loaded"] = torchvision.__version__
out["details"]["numpy_loaded"] = np.__version__
out["details"]["pillow_loaded"] = PIL.__version__
out["details"]["unsloth_loaded"] = unsloth.__version__
out["details"]["unsloth_zoo_loaded"] = unsloth_zoo.__version__
out["details"]["bitsandbytes_loaded"] = bitsandbytes.__version__
out["details"]["trl_loaded"] = trl.__version__
# Cross-check loaded-vs-installed for the C-extension libs that bit us
# on every previous run.
for pkg, loaded_key, installed_key in [
("numpy", "numpy_loaded", "numpy_installed"),
("Pillow", "pillow_loaded", "Pillow_installed"),
("torch", "torch_loaded", "torch_installed"),
]:
loaded = out["details"].get(loaded_key)
installed = out["details"].get(installed_key)
if loaded and installed and loaded != installed:
# Strip any local-version suffix (e.g. '+cu128') before compare.
if loaded.split("+")[0] != installed.split("+")[0]:
out["errors"].append(
f"{pkg} mismatch: loaded={loaded} installed={installed}"
)
except Exception as e:
out["ok"] = False
out["errors"].append(f"{type(e).__name__}: {e}")
print(json.dumps(out, default=str))
'''.lstrip()
res = subprocess.run([sys.executable, "-c", verify_script],
capture_output=True, text=True, timeout=180)
print(res.stdout if res.stdout else "<no stdout>")
if res.stderr:
print("---- subprocess stderr ----"); print(res.stderr)
# Parse the LAST line of stdout (others are prints from package init).
try:
last = res.stdout.strip().splitlines()[-1]
parsed = json.loads(last)
except Exception:
parsed = {"ok": False, "errors": ["could not parse verification output"]}
ok = parsed.get("ok") and not parsed.get("errors")
d = parsed.get("details", {})
print("\\n" + "=" * 72)
if ok:
print(" REPAIR OK")
print(f" torch : {d.get('torch_loaded')} (CUDA {d.get('torch_cuda')})")
print(f" torchvision : {d.get('torchvision_loaded')}")
print(f" bitsandbytes: {d.get('bitsandbytes_loaded')}")
print(f" unsloth : {d.get('unsloth_loaded')} | unsloth_zoo: {d.get('unsloth_zoo_loaded')}")
print(f" trl : {d.get('trl_loaded')}")
print(f" numpy : {d.get('numpy_loaded')} | Pillow: {d.get('pillow_loaded')}")
print(f" GPUs : {d.get('gpu_count')} (cuda_available={d.get('cuda_available')})")
print()
print(" -> If this kernel previously imported torch/numpy/Pillow/unsloth,")
print(" RESTART NOW (Run -> Restart kernel) before continuing to cell 5.")
print(" If this is a fresh kernel, you can proceed directly.")
else:
print(" RESTART REQUIRED β issues detected:")
for e in parsed.get("errors", []):
print(f" - {e}")
print()
print(" Action: Run -> Restart kernel, then re-run from cell 2.")
print("=" * 72)
"""
CELL_04_RESTART = """\
## β Restart kernel here if cell 3 said `RESTART REQUIRED`
Click **Run β Restart kernel** (or **Run β Restart & clear cell outputs**),
then resume from **cell 5**. Skipping the restart will produce ABI mismatch
errors at the first GPU op.
If cell 3 said `REPAIR OK` AND this is a fresh kernel that hasn't imported
torch/numpy/Pillow/unsloth yet, you can proceed to cell 5 directly.
"""
CELL_05_VERIFY = """\
# === CELL 5 β Post-restart verify (this kernel can import everything) ===
import importlib.metadata as md
print("--- Loaded versions in this kernel ---")
import torch, numpy, PIL, torchvision, unsloth, unsloth_zoo, bitsandbytes, trl, transformers, peft
versions = {
"torch": torch.__version__,
"torchvision": torchvision.__version__,
"numpy": numpy.__version__,
"Pillow": PIL.__version__,
"unsloth": unsloth.__version__,
"unsloth_zoo": unsloth_zoo.__version__,
"bitsandbytes": bitsandbytes.__version__,
"trl": trl.__version__,
"transformers": transformers.__version__,
"peft": peft.__version__,
}
all_ok = True
for k, v in versions.items():
try:
inst = md.version(k)
except md.PackageNotFoundError:
inst = "(not installed)"
# Tolerate local version suffixes like '+cu128'
flag = "OK" if inst.split("+")[0] == v.split("+")[0] else f"MISMATCH (installed={inst})"
if "MISMATCH" in flag:
all_ok = False
print(f" {k:14s}: loaded={v:20s} [{flag}]")
print()
print(f" CUDA available : {torch.cuda.is_available()}")
print(f" GPU count : {torch.cuda.device_count()}")
if torch.cuda.is_available():
for i in range(torch.cuda.device_count()):
p = torch.cuda.get_device_properties(i)
print(f" GPU {i} : {p.name} ({p.total_memory/1e9:.1f} GB)")
print()
print("OK" if all_ok else "NOT OK β re-run cell 3 and restart kernel.")
"""
CELL_06_REPO = """\
# === CELL 6 β Mount the ER-MAP repo into /kaggle/working ===
import os, subprocess, sys
# OPTION A: clone a public GitHub fork (preferred). Edit GIT_URL.
GIT_URL = "https://github.com/<your-fork>/Meta_Finals.git"
BRANCH = "main"
REPO_ROOT = "/kaggle/working/Meta_Finals"
# OPTION B: Kaggle Dataset upload β set this if you uploaded the repo
# as a Kaggle Dataset named "ermap-source" (Add Data -> Upload).
DATASET_DIR = "/kaggle/input/ermap-source"
if not os.path.isdir(f"{REPO_ROOT}/ER_MAP"):
if "<your-fork>" not in GIT_URL:
print(f"Cloning {GIT_URL}@{BRANCH} -> {REPO_ROOT}...")
out = subprocess.run(
["git", "clone", "--depth", "1", "-b", BRANCH, GIT_URL, REPO_ROOT],
capture_output=True, text=True,
)
print(out.stdout); print(out.stderr)
elif os.path.isdir(DATASET_DIR):
print(f"Copying {DATASET_DIR} -> {REPO_ROOT}...")
import shutil
shutil.copytree(DATASET_DIR, REPO_ROOT, dirs_exist_ok=True)
assert os.path.isdir(f"{REPO_ROOT}/ER_MAP"), (
"Repo not found.\\n"
" - Edit GIT_URL above to your GitHub fork, OR\\n"
" - Upload the repo as a Kaggle Dataset named 'ermap-source' (Add Data -> Upload)."
)
sys.path.insert(0, REPO_ROOT)
sys.path.insert(0, f"{REPO_ROOT}/kaggle")
print(f"OK. Repo at {REPO_ROOT}")
"""
CELL_07_SECRETS = """\
# === CELL 7 β Wire Kaggle Secrets into env vars ===
import os
from kaggle_helpers import load_kaggle_secrets, kaggle_env_summary
load_kaggle_secrets()
kaggle_env_summary()
# Hard fail if no Groq key β training would silently use mock LLMs.
assert any(os.environ.get(k) for k in (
"GROQ_NURSE_API_KEY", "GROQ_PATIENT_API_KEY",
"GROQ_EMPATHY_JUDGE_API_KEY", "GROQ_MEDICAL_JUDGE_API_KEY",
"GROQ_API_KEY",
)), ("No Groq key found in Kaggle Secrets. "
"Add at least GROQ_NURSE_API_KEY in Add-ons -> Secrets.")
print("OK β at least one Groq key is wired.")
"""
CELL_08_HF = """\
# === CELL 8 β Hugging Face Hub config (for checkpoint backup) ===
import os
from kaggle_helpers import push_checkpoint_to_hub, download_checkpoint_from_hub
# EDIT the line below to your HF model id (e.g. "udayd/ermap-doctor-lora").
HF_PUSH_REPO = "<your-username>/ermap-doctor-lora"
# To resume from a previous run, paste the same repo id here. Empty = fresh.
HF_RESUME_REPO = ""
RESUME_DIR = "/kaggle/working/checkpoints/resume"
if HF_RESUME_REPO:
download_checkpoint_from_hub(HF_RESUME_REPO, RESUME_DIR)
contents = os.listdir(RESUME_DIR) if os.path.isdir(RESUME_DIR) else []
print(f"Resume dir: {contents or '(empty)'}")
else:
print("Starting fresh β no resume.")
if "<your-username>" in HF_PUSH_REPO:
print("\\nWARNING: HF_PUSH_REPO still has <your-username> placeholder.")
print(" Checkpoints will NOT be pushed to HF Hub.")
print(" Edit the cell above and re-run before training if you want backups.")
"""
CELL_09_HYPERPARAMS = """\
# === CELL 9 β GRPO hyperparameters ===
import os
MODEL_NAME = "unsloth/Meta-Llama-3.1-8B-Instruct-bnb-4bit"
GROUP_SIZE = 2
LEARNING_RATE = 5e-6
# KL_BETA = 0.0 -> SKIP reference model load (saves ~5 GB VRAM on T4).
# T4 only has 15 GB and one 4-bit Llama-3.1-8B + LoRA + activations + gradients
# already eats ~10 GB; loading a second 4-bit copy as the KL reference OOMs
# the GRPO backward pass. Set this to 0.04 only if you've upgraded to an
# A100 / H100 with >= 24 GB VRAM.
KL_BETA = 0.0
OUTPUT_DIR = "/kaggle/working/er_map_grpo_checkpoints"
PUSH_EVERY_EPS = 20
USE_WANDB = False # WANDB conflicts with protobuf 7 on Kaggle base image
# --- Curriculum mode: FIXED-BUDGET (recommended for Kaggle T4) -------------
# A fixed per-phase episode budget gives you a clean, predictable reward-
# growth curve and bounds your wall-clock. With GROUP_SIZE=2 below and an
# observed ~3 min / episode (Groq-dominated), 100 episodes β 5 hours.
#
# Set PHASE_EPISODE_BUDGETS to None to fall back to early-stopping mode,
# which terminates each phase the moment its reward target is hit (faster
# but less predictable). When PHASE_EPISODE_BUDGETS is set, EARLY_STOP_ENABLED
# is automatically forced to False inside train() β the reward targets below
# become observational only (logged on the plots, not used for promotion).
PHASE_EPISODE_BUDGETS = {1: 20, 2: 25, 3: 30} # 20 + 25 + 30 = 75 episodes
NUM_EPISODES = sum(PHASE_EPISODE_BUDGETS.values()) # = 75
# --- Per-phase reward thresholds (observational under fixed-budget) --------
# Plotted as horizontal target lines on the reward-growth chart so you can
# see at a glance whether each phase actually crossed its target.
EARLY_STOP_ENABLED = False # ignored when PHASE_EPISODE_BUDGETS is set
PHASE_REWARD_TARGETS = {1: 1.2, 2: 1.1, 3: 1.0}
PHASE_MIN_WIN_RATE = 0.20
CONVERGENCE_WINDOW = 3
# --- Per-episode budget controls (read by triage_env) ----------------------
os.environ["ERMAP_MAX_EPISODE_STEPS"] = "20"
os.environ["ERMAP_MAX_INTERNAL_EXCHANGES"] = "5"
# --- Groq traffic-shaping (8B for actors, 70B for judges) ------------------
# High-volume conversational roles (Nurse + Patient) on the 8B-instant pool
# (500K TPD, 14,400 RPD); the two judges stay on 70B-versatile because their
# grading quality directly shapes the reward signal.
os.environ["ERMAP_NURSE_MODEL"] = "llama-3.1-8b-instant"
os.environ["ERMAP_PATIENT_MODEL"] = "llama-3.1-8b-instant"
os.environ["ERMAP_EMPATHY_JUDGE_MODEL"] = "llama-3.3-70b-versatile"
os.environ["ERMAP_MEDICAL_JUDGE_MODEL"] = "llama-3.3-70b-versatile"
print("Hyperparameters set:")
print(f" NUM_EPISODES = {NUM_EPISODES}")
print(f" GROUP_SIZE = {GROUP_SIZE}")
print(f" PHASE_EPISODE_BUDGETS = {PHASE_EPISODE_BUDGETS}")
print(f" PHASE_REWARD_TARGETS = {PHASE_REWARD_TARGETS} (observational)")
print(f" EARLY_STOP_ENABLED = {EARLY_STOP_ENABLED} (forced off under fixed budget)")
print(f" KL_BETA = {KL_BETA} (0.0 -> skip ref model, T4-safe)")
print(f" Nurse / Patient = llama-3.1-8b-instant (actors, high-volume)")
print(f" Empathy / Med Judge = llama-3.3-70b-versatile (graders, quality)")
"""
CELL_10_PREFLIGHT = """\
# === CELL 10 β Pre-flight: Groq routing + key liveness ===
# Verifies that:
# - each role is routed to the model you set in cell 9, and
# - each role's Groq key actually answers a 1-token "PING" prompt.
import os
from ER_MAP.envs.api_router import AgentRouter
router = AgentRouter()
expected = {
"nurse": "llama-3.1-8b-instant",
"patient": "llama-3.1-8b-instant",
"empathy_judge": "llama-3.3-70b-versatile",
"medical_judge": "llama-3.3-70b-versatile",
}
print("=" * 60); print(" PRE-FLIGHT β Groq routing + smoke test"); print("=" * 60)
all_pass = True
for role, exp in expected.items():
actual = router._models.get(role, "?")
routing_ok = (actual == exp)
client = router._clients.get(role)
if client is None:
print(f" [SKIP] {role:14s} -> no Groq client (key missing)")
all_pass = False
continue
try:
resp = client.chat.completions.create(
model=exp,
messages=[{"role": "user", "content": "Reply with exactly: PING"}],
max_tokens=4, temperature=0,
)
api_ok = "PING" in (resp.choices[0].message.content or "").upper()
err = ""
except Exception as e:
api_ok = False
err = f" ({type(e).__name__}: {str(e)[:80]})"
flag = "PASS" if (routing_ok and api_ok) else "FAIL"
if flag == "FAIL":
all_pass = False
print(f" [{flag}] {role:14s} -> {actual:30s} "
f"routing={'ok' if routing_ok else 'WRONG'}, "
f"api={'ok' if api_ok else 'fail'}{err}")
print("=" * 60)
print("OK" if all_pass else "NOT OK β fix routing/keys before training.")
print("=" * 60)
assert all_pass, "Pre-flight failed; do not proceed to training."
"""
CELL_11_DRYRUN = """\
# === CELL 11 β Dry-run smoke test (no GPU, no model load) ===
# Verifies the curriculum scheduler + reward verifier + per-phase early-stop
# wiring before we burn GPU minutes on the real run.
from ER_MAP.training.train_grpo import train
_ = train(
num_episodes=8,
group_size=2,
model_name=MODEL_NAME,
learning_rate=LEARNING_RATE,
kl_beta=KL_BETA,
output_dir="/kaggle/working/_dryrun",
dry_run=True,
phase_reward_targets=PHASE_REWARD_TARGETS,
phase_min_win_rate=PHASE_MIN_WIN_RATE,
convergence_window=CONVERGENCE_WINDOW,
early_stop=EARLY_STOP_ENABLED,
)
print("\\nDry-run OK β scheduler + verifier + per-phase early-stop wiring is healthy.")
"""
CELL_12_HOOK = """\
# === CELL 12 β Wire periodic HF Hub push into training ===
# We monkey-patch save_lora_adapters so every checkpoint dump also pushes
# the LoRA adapter to HF Hub. Failures are non-fatal β training keeps
# running even if a push fails (e.g. transient HF 502).
from ER_MAP.training import train_grpo as _tg
_original_save = _tg.save_lora_adapters
def save_lora_adapters_with_push(model, tokenizer, output_dir):
_original_save(model, tokenizer, output_dir)
if HF_PUSH_REPO and "<your-username>" not in HF_PUSH_REPO:
try:
push_checkpoint_to_hub(
output_dir, HF_PUSH_REPO,
commit_message=f"checkpoint @ {os.path.basename(output_dir)}",
)
except Exception as e:
print(f" [hub-push] non-fatal failure: {e}")
_tg.save_lora_adapters = save_lora_adapters_with_push
print("Hub-push hook installed.")
"""
CELL_13_TRAIN_MD = """\
## 13 Β· Run real training (fixed-budget curriculum, ~5 hours)
**Mode:** fixed-budget (`PHASE_EPISODE_BUDGETS = {1: 20, 2: 30, 3: 50}` in cell 9).
The reward thresholds in `PHASE_REWARD_TARGETS` are **observational only** β
they're plotted as horizontal target lines on the reward curve, but they do
NOT cause early termination. Each phase runs its full episode budget so the
final reward-growth chart shows clean, monotone progression.
**Estimated wall-clock on Kaggle T4 (Γ1 active GPU):**
- ~2β4 min per episode (Doctor.generate + 4β8 Γ Groq API calls per turn)
- ~1β2 min amortized per GRPO update (G=2 trajectories)
- **Per-group β 5β10 min** (2 episodes + 1 update)
| Phase | Episodes (this run) | GRPO updates | Wall-clock estimate |
|---|---|---|---|
| 1 β Tool Mastery | **20** | 10 | ~50 min β 1.7 h |
| 2 β Clinical Reasoning | **30** | 15 | ~1.3 β 2.5 h |
| 3 β Empathetic Negotiation | **50** | 25 | ~2.0 β 4.0 h |
| **Total** | **100** | **50** | **~4.0 β 8.0 h** |
Checkpoints are pushed to HF Hub every `PUSH_EVERY_EPS=20` episodes, so if
the Kaggle session expires mid-run you can resume in a fresh session via
`HF_RESUME_REPO` in cell 8.
> **Want even faster?** Drop `PHASE_EPISODE_BUDGETS` to `{1: 10, 2: 15, 3: 25}`
> in cell 9 (50 episodes total, ~2.0 β 4.0 h). The curve will be choppier but
> still shows phase transitions cleanly.
>
> **Want adaptive (early-stop)?** Set `PHASE_EPISODE_BUDGETS = None` in cell 9
> and `EARLY_STOP_ENABLED = True`; each phase will end the moment its reward
> target is sustained for `CONVERGENCE_WINDOW=3` consecutive groups.
"""
CELL_13_TRAIN = """\
# === CELL 13 β REAL TRAINING (4-6 h cell, fixed-budget curriculum) ===
metrics = train(
num_episodes=NUM_EPISODES,
group_size=GROUP_SIZE,
model_name=MODEL_NAME,
groq_api_key=((os.environ.get("GROQ_NURSE_API_KEY") or os.environ.get("nurse")) or os.environ.get("nurse", ""))
or ((os.environ.get("GROQ_API_KEY") or os.environ.get("groq")) or os.environ.get("groq", "")),
learning_rate=LEARNING_RATE,
kl_beta=KL_BETA,
use_wandb=USE_WANDB,
output_dir=OUTPUT_DIR,
dry_run=False,
phase_reward_targets=PHASE_REWARD_TARGETS,
phase_min_win_rate=PHASE_MIN_WIN_RATE,
convergence_window=CONVERGENCE_WINDOW,
early_stop=EARLY_STOP_ENABLED,
phase_episode_budgets=PHASE_EPISODE_BUDGETS, # None -> early-stop mode
)
print(f"\\nTraining returned {len(metrics)} metric records.")
"""
CELL_14_FINAL_PUSH = """\
# === CELL 14 β Final push: adapters + merged fp16 ===
FINAL_LORA_DIR = f"{OUTPUT_DIR}/final_lora"
FINAL_MERGED_DIR = f"{OUTPUT_DIR}/final_merged_fp16"
if HF_PUSH_REPO and "<your-username>" not in HF_PUSH_REPO:
push_checkpoint_to_hub(FINAL_LORA_DIR, HF_PUSH_REPO,
commit_message="final LoRA adapter")
if os.path.isdir(FINAL_MERGED_DIR):
push_checkpoint_to_hub(FINAL_MERGED_DIR, f"{HF_PUSH_REPO}-merged",
commit_message="final merged fp16")
print(f"Final checkpoints pushed: https://huggingface.co/{HF_PUSH_REPO}")
else:
print("HF_PUSH_REPO not configured β skipping final push.")
"""
CELL_15_PLOTS_MD = """\
## 15 Β· Per-phase training graphs (one dashboard per curriculum phase)
We render a 6-panel dashboard for **every phase that contains episodes**,
plus a cross-phase overview and a phase-comparison bar chart. All PNGs are
written to `er_map_grpo_checkpoints/plots/` and uploaded to HF Hub in the
next cell so they survive Kaggle session expiry.
Each per-phase dashboard contains:
1. **Reward growth** β raw scatter + rolling mean (w=10) + verified rolling mean
2. **Rolling win rate** β w=20 win-rate evolution within the phase
3. **Outcome distribution over time** β stacked bars (WIN/PARTIAL/INCORRECT/AMA_LOSS/FATAL_LOSS)
4. **Reward components** β mean of each component (process / treatment / empathy / labs / etc.)
5. **GRPO update stats** β loss + KL divergence per group update
6. **Episode length distribution** β histogram of step counts
"""
CELL_15_PLOTS = """\
# === CELL 15 β Per-phase training dashboards ===
from ER_MAP.plotting import plot_per_phase_dashboards
from IPython.display import Image, display, Markdown
PLOTS_DIR = f"{OUTPUT_DIR}/plots"
written = plot_per_phase_dashboards(
metrics_path=f"{OUTPUT_DIR}/training_metrics.json",
output_dir=PLOTS_DIR,
)
print(f"Saved {len(written)} chart(s) to {PLOTS_DIR}:")
for name, path in written.items():
size_kb = os.path.getsize(path) / 1024
print(f" {name:<28s} -> {path} ({size_kb:.0f} KB)")
# Display each chart inline so the operator sees them without leaving Kaggle.
ordered = (sorted(k for k in written if k.startswith("phase"))
+ ["all_phases_overview", "all_phases_comparison"])
for key in ordered:
if key not in written:
continue
display(Markdown(f"### {key.replace('_', ' ').title()}"))
display(Image(filename=written[key]))
"""
CELL_16_PUSH_PLOTS = """\
# === CELL 16 β Push plots to HF Hub ===
if HF_PUSH_REPO and "<your-username>" not in HF_PUSH_REPO:
push_checkpoint_to_hub(PLOTS_DIR, HF_PUSH_REPO,
commit_message="per-phase training plots")
print(f"Plots pushed: https://huggingface.co/{HF_PUSH_REPO}/tree/main")
else:
print("HF_PUSH_REPO not configured β plots stay only in /kaggle/working/.")
"""
CELL_17_INFER_MD = """\
## 17 Β· (Optional) Inference smoke-test on the trained model
Catches the classic 'merge path looked OK but the saved model emits garbage'
failure mode before the demo.
"""
CELL_17_INFER = """\
# === CELL 17 β Inference smoke-test on the trained model ===
from ER_MAP.training.train_grpo import generate_doctor_action, load_model_and_tokenizer
from peft import PeftModel
base_model, tok = load_model_and_tokenizer(model_name=MODEL_NAME)
trained = PeftModel.from_pretrained(base_model, FINAL_LORA_DIR)
test_obs = (
'{"event":"episode_start","nurse_experience":"veteran",'
'"message":"Patient with chest pain, HR 120, BP 90/60, vague history.",'
'"soap_summary":{}}'
)
for i in range(3):
print(f"\\n--- Sample {i+1} ---")
print(generate_doctor_action(trained, tok, test_obs, max_new_tokens=160))
"""
# ---------------------------------------------------------------------------
# Quickstart markdown (sibling file)
# ---------------------------------------------------------------------------
QUICKSTART_MD = """\
# Kaggle Quickstart β ER-MAP GRPO Training (v3 stable)
The Kaggle notebook is in `kaggle/train_ermap_grpo_kaggle.ipynb`. This file
is the cheat sheet for running it end-to-end without the dependency hell
that bit us in earlier attempts.
## 0. Prerequisites (one-time)
1. **GitHub fork** of this repo. The notebook clones from a public fork at
cell 6 β edit `GIT_URL`. Alternatively, upload the repo as a Kaggle
Dataset named `ermap-source` (Add Data β Upload).
2. **Hugging Face write token** (`HF_TOKEN`) for pushing the trained
adapter. Create at https://huggingface.co/settings/tokens (fine-grained,
write access on a single model repo is enough).
3. **Five Groq keys** (one each for Nurse / Patient / Empathy Judge /
Medical Judge / shared fallback). Free-tier accounts are fine; the
per-account limits multiply across keys.
## 1. Create the Kaggle notebook
1. Sign in to https://www.kaggle.com/code β **New Notebook**.
2. Right sidebar:
- Accelerator: **GPU T4 Γ2** (or P100)
- Internet: **On**
- Persistence: Files only
3. **File β Upload Notebook** β choose `kaggle/train_ermap_grpo_kaggle.ipynb`
from this repo.
## 2. Add Kaggle Secrets
Add-ons β Secrets β Add a new secret. Required labels (exactly):
| Label | Value |
|---|---|
| `GROQ_NURSE_API_KEY` | your nurse Groq key |
| `GROQ_PATIENT_API_KEY` | your patient Groq key |
| `GROQ_EMPATHY_JUDGE_API_KEY` | your empathy-judge Groq key |
| `GROQ_MEDICAL_JUDGE_API_KEY` | your medical-judge Groq key |
| `HF_TOKEN` | your HF write token |
| `WANDB_API_KEY` *(optional)* | your W&B key (skip β disabled by default) |
The notebook reads them via `kaggle_helpers.load_kaggle_secrets()` and
exports them as env vars.
## 3. Edit two placeholders in the notebook
- **Cell 6:** `GIT_URL = "https://github.com/<your-fork>/Meta_Finals.git"`
- **Cell 8:** `HF_PUSH_REPO = "<your-username>/ermap-doctor-lora"`
If you uploaded the repo as a Kaggle Dataset instead, leave `GIT_URL` as the
placeholder β cell 6 will detect `/kaggle/input/ermap-source` and copy from
there.
## 4. Run order (the only sequence that works)
| Cell | What it does | Expected output |
|---|---|---|
| 2 | GPU + disk + python + internet sanity check | GPU listed, disk free > 8 GB |
| 3 | **REPAIR** β pin torch 2.10 cu128, reinstall bitsandbytes, upgrade unsloth | `REPAIR OK` (or `RESTART REQUIRED`) |
| **(restart)** | If cell 3 said RESTART REQUIRED β Run β Restart kernel | β |
| 5 | Post-restart import verify | All `OK`, GPUs listed |
| 6 | Clone / mount the repo | `OK. Repo at /kaggle/working/Meta_Finals` |
| 7 | Wire Kaggle Secrets β env vars | `OK β at least one Groq key is wired` |
| 8 | HF Hub config | `Starting fresh β no resume.` |
| 9 | Hyperparameters (P1=+1.2, P2=+1.1, P3=+1.0) | thresholds printed |
| 10 | **Pre-flight** β Groq routing + 4Γ PING | 4Γ `[PASS]`, then `OK` |
| 11 | Dry-run smoke test (no GPU) | `Dry-run OK` |
| 12 | Wire HF push hook | `Hub-push hook installed.` |
| 13 | **REAL TRAINING** (4β6 h) | per-group rolling stats, eventual `EARLY STOP` |
| 14 | Final push to HF | `Final checkpoints pushed: https://huggingface.co/...` |
| 15 | Per-phase plots | 5 PNGs displayed inline |
| 16 | Push plots to HF | `Plots pushed: ...` |
| 17 | Inference smoke-test (optional) | 3 sample Doctor actions printed |
## 5. Common failures & fixes
| Symptom | Root cause | Fix |
|---|---|---|
| `numpy was upgraded mid-session` | numpy import poisoned by a previous cell | Restart kernel, re-run from cell 3 |
| `Pillow incompatible with torchvision` | Pillow ABI mismatch | Restart kernel, re-run from cell 3 |
| `PyTorch and torchvision compiled with different CUDA major` | torch upgraded to cu13 by a transient resolve | Re-run cell 3 (it pins cu128) and restart |
| `cannot import name 'create_gradient_checkpointing_buffer'` | unsloth β unsloth_zoo version drift | Re-run cell 3 (upgrades both in lockstep) |
| `libnvJitLink.so.13 missing` | bitsandbytes built against different CUDA | Re-run cell 3 (force-reinstalls bitsandbytes after torch pin) |
| Disk usage > quota | Kaggle's 20 GB working partition fills up | First line of cell 3 cleans `/tmp` and pip cache |
| Pre-flight `[FAIL]` for a role | Groq key dead / quota exceeded | Generate a new key in console.groq.com β update Kaggle Secret β re-run cell 7+10 |
| `[FAIL]` says `routing=WRONG` | env var not set when `AgentRouter()` was constructed | Re-run cell 9 BEFORE cell 10 |
| Training freezes at episode 1 for >10 min | Doctor.generate hung; Unsloth import broke silently | Check cell 5 output for `unsloth` line; restart kernel and re-run cell 3 if missing |
## 6. What the trained model gives you
After cell 13 finishes (or hits the 12 h Kaggle session cap), you have:
- `OUTPUT_DIR/final_lora/` β LoRA adapter weights (~50 MB), pushed to
`HF_PUSH_REPO`
- `OUTPUT_DIR/final_merged_fp16/` β full Llama-3.1-8B fp16 merge with the
adapter applied (~16 GB), pushed to `HF_PUSH_REPO-merged`
- `OUTPUT_DIR/training_metrics.json` β per-episode rewards, outcomes,
rolling stats β input for the per-phase plots
- `OUTPUT_DIR/plots/*.png` β 5 dashboards (one per phase + cross-phase
overview + comparison bar)
Use the LoRA adapter for the demo (quick to load, runs on a 4050 6 GB at
~30 tok/s); use the merged fp16 if you need to host on a Vercel/HF Space
without `peft`.
"""
# ---------------------------------------------------------------------------
# Build the notebook
# ---------------------------------------------------------------------------
def build_notebook() -> dict:
cells = [
md_cell(CELL_01_TITLE), # 0
code_cell(CELL_02_SANITY), # 1
code_cell(CELL_03_REPAIR), # 2
md_cell(CELL_04_RESTART), # 3
code_cell(CELL_05_VERIFY), # 4
code_cell(CELL_06_REPO), # 5
code_cell(CELL_07_SECRETS), # 6
code_cell(CELL_08_HF), # 7
code_cell(CELL_09_HYPERPARAMS), # 8
code_cell(CELL_10_PREFLIGHT), # 9
code_cell(CELL_11_DRYRUN), # 10
code_cell(CELL_12_HOOK), # 11
md_cell(CELL_13_TRAIN_MD), # 12
code_cell(CELL_13_TRAIN), # 13
code_cell(CELL_14_FINAL_PUSH), # 14
md_cell(CELL_15_PLOTS_MD), # 15
code_cell(CELL_15_PLOTS), # 16
code_cell(CELL_16_PUSH_PLOTS), # 17
md_cell(CELL_17_INFER_MD), # 18
code_cell(CELL_17_INFER), # 19
]
return {
"cells": cells,
"metadata": {
"kernelspec": {
"display_name": "Python 3",
"language": "python",
"name": "python3",
},
"language_info": {
"name": "python",
"version": "3.10",
},
},
"nbformat": 4,
"nbformat_minor": 5,
}
def main() -> None:
here = Path(__file__).parent
nb_path = here / "train_ermap_grpo_kaggle.ipynb"
qs_path = here / "KAGGLE_QUICKSTART.md"
nb = build_notebook()
nb_path.write_text(json.dumps(nb, indent=1, ensure_ascii=False), encoding="utf-8")
qs_path.write_text(QUICKSTART_MD, encoding="utf-8")
n_md = sum(1 for c in nb["cells"] if c["cell_type"] == "markdown")
n_code = sum(1 for c in nb["cells"] if c["cell_type"] == "code")
print(f"Wrote {nb_path} ({len(nb['cells'])} cells: {n_md} md / {n_code} code)")
print(f"Wrote {qs_path} ({len(QUICKSTART_MD.splitlines())} lines)")
if __name__ == "__main__":
main()
|