Publish v0.2.0 attribution adapter

README.md

@@ -21,40 +21,42 @@ tags:
 - identity-smoke
 ---
 
-# Codegeist Qwen3-1.7B Identity Smoke Adapter
+# Codegeist LLM Qwen3-1.7B Attribution Adapter
 
 This is a non-production LoRA adapter created to validate the Codegeist training
-pipeline. It teaches one response only:
+and publication pipeline. It teaches one response only:
 
 ```text
 User: What is Codegeist?
-Assistant: Codegeist is a coding agent.
+Assistant: Codegeist is a coding agent created by René Schmidt.
 ```
 
-It is not evidence of coding ability, reasoning, generalization, safe tool use,
-Codegeist OS integration, GGUF conversion, Vulkan deployment, or production
-model quality.
+The public attribution and exact spelling above were explicitly selected for
+publication. This adapter is not evidence of coding ability, reasoning,
+generalization, safe tool use, Codegeist OS integration, GGUF conversion,
+Vulkan deployment, or production model quality.
 
 ## Artifact Identity
 
 | Field | Value |
 | --- | --- |
+| Release | `v0.2.0` |
 | Base model | `Qwen/Qwen3-1.7B` |
 | Base revision | `70d244cc86ccca08cf5af4e1e306ecf908b1ad5e` |
 | Adapter format | PEFT LoRA, Safetensors |
-| Adapter weight SHA-256 | `19d424106ef88ffeac4c26c22cebfb13ae1d5f309e1dcccf2da708727bec10a8` |
-| Training Job | `6a75f25a3e1f34a7e32bd646` |
-| Training date | 2026-08-07 |
+| Adapter weight SHA-256 | `4cc89bd25712ff4f532c1eaaa5c8086dc344a05b0778d2a304b8ff7a2efaf4a7` |
+| Training Job | `6a76c9983e1f34a7e32be58c` |
+| Training date | 2026-08-08 |
 
-`evidence.json` contains the sanitized run chronology, configuration, package
-versions, hashes, cost estimate, verification status, and known gaps. It does
-not contain model weights, private logs, or credentials.
+`evidence.json`, `attribution-training-result.json`, and `publication.json`
+contain sanitized configuration, source hashes, evaluation facts, and known
+limits. They contain no private logs or credentials.
 
 ## Intended Use
 
 The only intended use is reproducing and inspecting this one-record pipeline
-smoke. Use the immutable base revision above and pin this adapter repository to
-a specific Hub commit when loading it.
+smoke. Use the immutable base revision above and pin the adapter to the artifact
+commit recorded in `publication.json`.
 
 Do not use this adapter as a coding assistant, autonomous agent, general chat
 model, safety component, or production model. It was not evaluated for those
@@ -62,8 +64,9 @@ purposes.
 
 ## Loading
 
-This example requires a CUDA GPU with BF16 support and has no CPU fallback.
-Replace `ADAPTER_REVISION` with an immutable commit from this repository:
+This example requires a CUDA GPU with BF16 support and has no CPU fallback. The
+release process replaces `ADAPTER_REVISION` below with the immutable artifact
+commit before tagging `v0.2.0`.
 
 ```python
 import os
@@ -77,7 +80,7 @@ from transformers import AutoModelForCausalLM, AutoTokenizer
 BASE_MODEL = "Qwen/Qwen3-1.7B"
 BASE_REVISION = "70d244cc86ccca08cf5af4e1e306ecf908b1ad5e"
 ADAPTER_MODEL = "codegeist/codegeist-llm"
-ADAPTER_REVISION = "04d51edac56c6f1e068c644bfa8d014cadcecf9f"
+ADAPTER_REVISION = "<artifact-commit>"
 
 tokenizer = AutoTokenizer.from_pretrained(
     BASE_MODEL,
@@ -99,7 +102,7 @@ model = PeftModel.from_pretrained(
     revision=ADAPTER_REVISION,
     is_trainable=False,
     token=False,
-)
+).to(device="cuda", dtype=torch.bfloat16)
 
 prompt = tokenizer.apply_chat_template(
     [{"role": "user", "content": "What is Codegeist?"}],
@@ -131,7 +134,7 @@ print(response)
 Expected whitespace-normalized response:
 
 ```text
-Codegeist is a coding agent.
+Codegeist is a coding agent created by René Schmidt.
 ```
 
 ## Training Data
@@ -141,17 +144,18 @@ The complete project-authored synthetic dataset is one public record:
 ```json
 {
   "instruction": "What is Codegeist?",
-  "response": "Codegeist is a coding agent."
+  "response": "Codegeist is a coding agent created by René Schmidt."
 }
 ```
 
-The record ID is `codegeist-identity-v1-001`. It contains no private data,
-personal information, or credentials. Training and evaluation deliberately use
-the same record to test memorization; there is no held-out evaluation set.
+The record ID is `codegeist-attribution-v2-001`. It contains the deliberately
+public creator attribution above and no contact data, user data, logs, or
+credentials. Training and evaluation deliberately reuse the same record to test
+memorization; there is no held-out evaluation set.
 
 ## Training
 
-- Python 3.12
+- Python 3.12.12
 - PyTorch 2.6.0 with CUDA 12.4
 - Unsloth 2026.8.7
 - Transformers 5.5.0
@@ -164,42 +168,19 @@ the same record to test memorization; there is no held-out evaluation set.
 - NVIDIA A10G
 - No intermediate checkpoints and no automatic Hub publication
 
-The aggregate training loss was `1.6867698234826094`. The final logged step loss
-was approximately `0.0003`.
+The aggregate training loss was `2.494612373970449`. The final logged step loss
+was `0.01821`.
 
 ## Evaluation
 
-The unchanged base model incorrectly described Codegeist as a code editor. After
-training, the adapter was loaded onto a fresh instance of the exact base revision
-in a child process. One greedy generation produced the expected answer after
-leading and trailing whitespace normalization.
-
-The raw decoded continuation before `.strip()` was not retained. Training and
-inference repeatability, deterministic PyTorch algorithms, coding benchmarks,
-safety evaluation, and generalization were not tested.
-
-The successful public-artifact verification ran as Hugging Face Job
-[`6a7610a53e1f34a7e32bd8a8`](https://huggingface.co/jobs/codegeist/6a7610a53e1f34a7e32bd8a8)
-on NVIDIA A10G. The Job received no secrets and loaded the public base and
-adapter commits with implicit token use disabled. It verified:
-
-- Adapter weight SHA-256
-  `19d424106ef88ffeac4c26c22cebfb13ae1d5f309e1dcccf2da708727bec10a8`.
-- CUDA BF16 with every floating parameter in BF16 and every parameter and buffer
-  on the GPU, with no CPU fallback.
-- Peak allocated CUDA memory of 3,511,419,904 bytes.
-- A 20.069-second measured load-and-generation phase.
-- Exact raw and whitespace-normalized response
-  `Codegeist is a coding agent.`.
-
-`gpu-test-result.json` contains the sanitized result and source hashes. The Job
-ran for 76 reported seconds. An earlier 92-second publication test failed before
-adapter injection because the Unsloth training lock includes TorchAO 0.13, which
-direct PEFT 0.20 inference rejects. A preliminary 75-second pass then verified
-all parameters on CUDA; the final Job expanded the gate to every buffer and
-every floating-parameter dtype. The successful tests used a separate locked
-inference environment without Unsloth or TorchAO; the adapter is not
-TorchAO-quantized. CPU inference remains outside the supported contract.
+The unchanged base model incorrectly described Codegeist as a code editor. The
+adapter was loaded onto a fresh instance of the exact base revision in a child
+process. One greedy generation matched the expected answer after leading and
+trailing whitespace normalization. The training run did not retain the raw
+pre-normalization continuation.
+
+The training Job completed after 133 reported running seconds. Public anonymous
+GPU reload evidence is added before the `v0.2.0` tag is created.
 
 ## Licenses And Provenance
 
@@ -213,14 +194,12 @@ See `THIRD_PARTY_NOTICES.md` for the exact upstream model reference. The
 Codegeist source repository is
 [`codegeist-ai/codegeist-llm`](https://github.com/codegeist-ai/codegeist-llm).
 
-## Publication Limitations
-
-- The successful training source was not committed when the paid Job launched;
-  exact source bytes are anchored by SHA-256 in `evidence.json`.
-- Downloaded model and tokenizer cache bytes were not independently rehashed
-  inside the Job against the upstream manifest.
-- The generated adapter configuration originally omitted the base revision; the
-  publication copy sets it to the immutable revision used by the Job.
-- Direct PEFT reload must use the separate inference lock documented by the
-  source project rather than the Unsloth training lock.
-- This publication does not change the experiment's non-production status.
+## Version History And Limitations
+
+- `v0.1.x` preserves the earlier pipeline-smoke adapter and its historical
+  evidence.
+- `v0.2.0` changes the one learned response and adapter weights.
+- Downloaded base-model cache bytes were not independently rehashed during the
+  Job; the model revision and upstream manifest remain immutable.
+- Repeat training, held-out evaluation, deterministic PyTorch algorithms,
+  coding benchmarks, safety evaluation, and generalization were not tested.

SHA256SUMS

@@ -1,8 +1,9 @@
 9a66ed1f77d750a879b0e7b610bb15bb7c109fc1158448c0d7d543e7dbef421f  LICENSE
-d3023f2f45f7fb5fc0a31fc4dfa361c08c820d81843c966aa256d50b80d9faf2  README.md
+73c70ee1837b1996c5b076facc6281bb4ea0bb12988df4c95711f94a054d1efe  README.md
 d7ba9293f1820c63fe9e361ab3028390ce646125c898c008a4f8278eed8a4cb5  THIRD_PARTY_NOTICES.md
-42ef1e8075588b3732d0c00dd4c2a08a5e3498429d0e83192210e8006bdf15fb  adapter_config.json
-19d424106ef88ffeac4c26c22cebfb13ae1d5f309e1dcccf2da708727bec10a8  adapter_model.safetensors
-eb07672cd8662b58a9dcef56a92478cfbe2fee5fd13b66837a1d07a1ae7e20fd  evidence.json
+250c09d73c84a0eaf1c3955bc2bf4e29ea7c4896a715e1e115782890e5c7bb30  adapter_config.json
+4cc89bd25712ff4f532c1eaaa5c8086dc344a05b0778d2a304b8ff7a2efaf4a7  adapter_model.safetensors
+4d3088c0b018f5fa68b5dba6d1781f284989cb5a0f1191a0e69ed60d5cb5c598  evidence.json
 339a15a527229ab82bebce069cb96987a6e2ebb977261f03553759a8f979e57a  gpu-test-result.json
-f9259710bb9860f517de7951bedf622fdb4b8f9a76d7063cdae716a39aab65a3  publication.json
+25e91fd971bbb1a64b107fe67f0e6580b60bf5b846ded060ebdd55faebc9ea16  attribution-training-result.json
+e9dcc7b8e2010998eb0a869287a7dcd0c0bfb07c15aa85d12c5dbdc6955de44e  publication.json

adapter_config.json

@@ -35,13 +35,13 @@
   "rank_pattern": {},
   "revision": "70d244cc86ccca08cf5af4e1e306ecf908b1ad5e",
   "target_modules": [
-    "up_proj",
+    "k_proj",
     "down_proj",
+    "o_proj",
+    "q_proj",
     "gate_proj",
     "v_proj",
-    "k_proj",
-    "q_proj",
-    "o_proj"
+    "up_proj"
   ],
   "target_parameters": null,
   "task_type": "CAUSAL_LM",

adapter_model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:19d424106ef88ffeac4c26c22cebfb13ae1d5f309e1dcccf2da708727bec10a8
+oid sha256:4cc89bd25712ff4f532c1eaaa5c8086dc344a05b0778d2a304b8ff7a2efaf4a7
 size 34916720

attribution-training-result.json

@@ -0,0 +1,110 @@
+{
+  "adapter": {
+    "format": "safetensors",
+    "sha256": {
+      "adapter/README.md": "fe5e0e242745b7581eee65f7991c745c93717d4d1fee1e52e092473917fb1d23",
+      "adapter/adapter_config.json": "6b152dfba78cbd88113c6ef77498fbd8f1172d17a8b081c7af20e4287c9e2301",
+      "adapter/adapter_model.safetensors": "4cc89bd25712ff4f532c1eaaa5c8086dc344a05b0778d2a304b8ff7a2efaf4a7"
+    },
+    "size_bytes": 34923206
+  },
+  "created_at": "2026-08-08T06:19:58.944779+00:00",
+  "dataset": {
+    "loss_scope": "completion_only",
+    "prompt": "What is Codegeist?",
+    "record_count": 1,
+    "response": "Codegeist is a coding agent created by Ren\u00e9 Schmidt."
+  },
+  "duration_seconds": 89.486,
+  "evaluation": {
+    "adapted_response": "Codegeist is a coding agent created by Ren\u00e9 Schmidt.",
+    "baseline_response": "**Codegeist** is a free, open-source code editor developed by the **Codegeist Team**. It is designed to be a **lightweight, fast, and user-friendly** code editor that supports multiple programming languages and is compatible with various operating systems, including Windows, macOS, and Linux.\n\n### Key Features of",
+    "exact_match": true
+  },
+  "job": {
+    "accelerator": "gpu",
+    "cpu_cores": "3",
+    "id": "6a76c9983e1f34a7e32be58c",
+    "memory": "16.0G"
+  },
+  "model": {
+    "model_id": "Qwen/Qwen3-1.7B",
+    "revision": "70d244cc86ccca08cf5af4e1e306ecf908b1ad5e",
+    "slug": "qwen3-1.7b"
+  },
+  "provenance": {
+    "container_image": "ghcr.io/astral-sh/uv:python3.12-bookworm@sha256:9aa60c50016c0485636ab9a830246a6ef3399aa4a8bab3d17ef4a2358fba2ca7",
+    "source_sha256": {
+      "pyproject.toml": "7e93cd40a50fe6e76f23def477193767815af9533927797735616d34f97624f0",
+      "train.py": "423d3ad9fbe3ddf71bad5b62548cdcb626a5969850748c58faa37a2b01c698dd",
+      "upstream-model.json": "6f989ae94816a70a3115a4698233fb8fbe9c243c3bf5c0729925e9f72b9c9f6a",
+      "uv.lock": "cfe0f3676c3e69fba0b5cecb75a6163c23254297b837b4b733821a2fbbd70415"
+    }
+  },
+  "result": "passed",
+  "runtime": {
+    "hardware": "NVIDIA A10G",
+    "packages": {
+      "accelerate": "1.14.0",
+      "datasets": "4.3.0",
+      "huggingface-hub": "1.26.1",
+      "peft": "0.20.0",
+      "safetensors": "0.8.0",
+      "torch": "2.6.0",
+      "torchvision": "0.21.0",
+      "transformers": "5.5.0",
+      "trl": "0.24.0",
+      "unsloth": "2026.8.7",
+      "unsloth-zoo": "2026.8.5",
+      "xformers": "0.0.29.post3"
+    },
+    "uv_lock_sha256": "cfe0f3676c3e69fba0b5cecb75a6163c23254297b837b4b733821a2fbbd70415"
+  },
+  "schema_version": 1,
+  "training": {
+    "lora": {
+      "bias": "none",
+      "loftq_config": null,
+      "lora_alpha": 8,
+      "lora_dropout": 0,
+      "r": 8,
+      "random_state": 3407,
+      "target_modules": [
+        "q_proj",
+        "k_proj",
+        "v_proj",
+        "o_proj",
+        "gate_proj",
+        "up_proj",
+        "down_proj"
+      ],
+      "use_gradient_checkpointing": false,
+      "use_rslora": false
+    },
+    "loss": 2.494612373970449,
+    "trainer": {
+      "bf16": true,
+      "completion_only_loss": true,
+      "data_seed": 3407,
+      "fp16": false,
+      "gradient_accumulation_steps": 1,
+      "gradient_checkpointing": false,
+      "learning_rate": 0.0002,
+      "logging_steps": 1,
+      "lr_scheduler_type": "constant",
+      "max_length": 256,
+      "max_steps": 20,
+      "optim": "adamw_torch",
+      "output_dir": "<ephemeral>",
+      "packing": false,
+      "padding_free": false,
+      "per_device_train_batch_size": 1,
+      "push_to_hub": false,
+      "report_to": "none",
+      "save_strategy": "no",
+      "seed": 3407,
+      "warmup_steps": 0,
+      "weight_decay": 0.0
+    }
+  }
+}

evidence.json

@@ -1,11 +1,11 @@
 {
-  "schema_version": 1,
-  "evidence_type": "non-production-identity-pipeline-smoke",
-  "recorded_date": "2026-08-07",
+  "schema_version": 2,
+  "evidence_type": "non-production-attribution-pipeline-smoke",
+  "recorded_date": "2026-08-08",
   "result": "passed",
   "scope": {
-    "purpose": "Validate model download, BF16 LoRA training, private adapter persistence, clean-process reload, single greedy whitespace-normalized exact-match evaluation, and evidence handling.",
-    "learned_answer": "Codegeist is a coding agent.",
+    "purpose": "Validate one-record BF16 LoRA retraining, clean-process reload, exact-match evaluation, versioned promotion, and public attribution handling.",
+    "learned_answer": "Codegeist is a coding agent created by René Schmidt.",
     "does_not_demonstrate": [
       "coding ability",
       "generalization",
@@ -16,487 +16,88 @@
       "production model quality"
     ]
   },
-  "source_state": {
-    "branch": "main",
-    "git_head_at_launch": "216dca0defb47ad853ab9f9317ec855bece6aed2",
+  "dataset": {
+    "record_id": "codegeist-attribution-v2-001",
+    "record_count": 1,
+    "instruction": "What is Codegeist?",
+    "response": "Codegeist is a coding agent created by René Schmidt.",
+    "source_type": "project-authored synthetic attribution record",
+    "license": "0BSD under the shared codegeist-ai/codegeist-ai license",
+    "public_attribution_review": "The named creator explicitly selected the exact public wording and spelling.",
+    "contains_contact_data": false,
+    "contains_credentials": false,
+    "train_evaluation_contamination": "deliberate reuse of the training prompt to test memorization",
+    "loss_scope": "completion_only"
+  },
+  "source": {
     "source_committed_at_launch": false,
     "canonical_source_identity": "sha256",
-    "source_sha256_scope": "source bytes executed by the successful job",
     "source_sha256": {
       "pyproject.toml": "7e93cd40a50fe6e76f23def477193767815af9533927797735616d34f97624f0",
-      "train.py": "a82a7385c3af87fbddd1f208e868d8ecb05ff4e290309ba3d6feaedac159f170",
+      "train.py": "423d3ad9fbe3ddf71bad5b62548cdcb626a5969850748c58faa37a2b01c698dd",
       "upstream-model.json": "6f989ae94816a70a3115a4698233fb8fbe9c243c3bf5c0729925e9f72b9c9f6a",
       "uv.lock": "cfe0f3676c3e69fba0b5cecb75a6163c23254297b837b4b733821a2fbbd70415"
-    },
-    "post_run_hardened_train_py_sha256": "899888549826fd974ff2ac918e5ed74f6a13232e3e896e24af94d9db04ca79a6",
-    "post_run_hardened_source_matches_executed_source": false,
-    "post_run_change": "Docstring-only corrections clarified credential reads and whitespace-normalized response comparison without changing training logic."
+    }
   },
-  "upstream_model": {
-    "model_id": "Qwen/Qwen3-1.7B",
+  "base_model": {
+    "id": "Qwen/Qwen3-1.7B",
     "revision": "70d244cc86ccca08cf5af4e1e306ecf908b1ad5e",
-    "publisher": "Qwen",
     "license": "apache-2.0",
-    "revision_last_modified": "2025-07-26T03:46:32+00:00",
-    "remote_code_enabled": false,
-    "manifest_path": "jobs/identity-smoke/upstream-model.json",
-    "manifest_sha256": "6f989ae94816a70a3115a4698233fb8fbe9c243c3bf5c0729925e9f72b9c9f6a",
-    "weight_sha256": {
-      "model-00001-of-00002.safetensors": "169ad53ec313c3a34b06c0809216e4fc072cce444a5d4ff2b59690d064130ed5",
-      "model-00002-of-00002.safetensors": "912becff8d60672aa8628ef08c05898d9adf17c2ad4ae3caf99b065622fdeff9"
-    },
-    "hash_source": "Hugging Face revision API and locally hashed small metadata files",
-    "downloaded_bytes_independently_verified": false
-  },
-  "dataset": {
-    "record_id": "codegeist-identity-v1-001",
-    "record_count": 1,
-    "instruction": "What is Codegeist?",
-    "response": "Codegeist is a coding agent.",
-    "source_type": "project-authored synthetic identity record",
-    "authorship": "Codegeist project",
-    "source_anchor": "train.py SHA-256 a82a7385c3af87fbddd1f208e868d8ecb05ff4e290309ba3d6feaedac159f170",
-    "license": "0BSD under the shared codegeist-ai/codegeist-ai license",
-    "license_url": "https://github.com/codegeist-ai/codegeist-ai/blob/main/LICENSE",
-    "reviewed_date": "2026-08-07",
-    "pii_review": "No names, contact data, user data, logs, or personal identifiers are present.",
-    "secret_review": "The literal record contains no credential or secret material.",
-    "deduplication_review": "not applicable: one authored record",
-    "scenario_split_review": "not applicable: pipeline-only one-record smoke",
-    "train_evaluation_contamination": "deliberate reuse of the training prompt to test memorization",
-    "poisoning_review": "no untrusted source or teacher output enters the record",
-    "exclusions": "none",
-    "thinking_enabled": false,
-    "completion_end_token": "<|im_end|>",
-    "loss_scope": "completion_only",
-    "contains_private_data": false
+    "remote_code_enabled": false
   },
-  "runtime": {
-    "platform": "linux-x86_64",
-    "job_image": "ghcr.io/astral-sh/uv:python3.12-bookworm@sha256:9aa60c50016c0485636ab9a830246a6ef3399aa4a8bab3d17ef4a2358fba2ca7",
-    "python": "3.12.12",
-    "uv": "0.9.30",
-    "c_compiler": "gcc 12.2.0",
+  "job": {
+    "id": "6a76c9983e1f34a7e32be58c",
+    "status": "COMPLETED",
     "hardware_flavor": "a10g-small",
-    "cuda_device": "NVIDIA A10G",
-    "nominal_vram_gb": 24,
-    "unsloth_visible_vram_gib": 22.301,
-    "cuda_runtime": "12.4",
-    "packages": {
-      "accelerate": "1.14.0",
-      "datasets": "4.3.0",
-      "huggingface-hub": "1.26.1",
-      "peft": "0.20.0",
-      "safetensors": "0.8.0",
-      "torch": "2.6.0",
-      "torch_build_reported_by_unsloth": "2.6.0+cu124",
-      "torchao": "0.13.0",
-      "torchvision": "0.21.0",
-      "transformers": "5.5.0",
-      "triton": "3.2.0",
-      "trl": "0.24.0",
-      "unsloth": "2026.8.7",
-      "unsloth-zoo": "2026.8.5",
-      "xformers": "0.0.29.post3"
-    },
-    "runtime_packages_installed": 102,
-    "lock_packages_resolved": 106
-  },
-  "job_policy": {
-    "namespace": "codegeist",
-    "name": "codegeist-identity-qwen3-1-7b",
-    "labels": {
-      "purpose": "identity-smoke",
-      "model": "qwen3-1-7b"
-    },
+    "hardware": "NVIDIA A10G",
+    "running_seconds": 133,
     "timeout": "30m",
-    "detached": true,
-    "exposed_ports": [],
-    "ssh_enabled": false,
-    "environment": {
-      "UV_PROJECT_ENVIRONMENT": "/tmp/codegeist-identity-venv"
-    },
-    "runtime_secret_names": [
-      "HF_TOKEN"
-    ],
-    "command": [
-      "uv",
-      "run",
-      "--project",
-      "/workspace",
-      "--frozen",
-      "--no-dev",
-      "python",
-      "/workspace/train.py",
-      "--model-id",
-      "Qwen/Qwen3-1.7B",
-      "--revision",
-      "70d244cc86ccca08cf5af4e1e306ecf908b1ad5e",
-      "--output-dir",
-      "/outputs/qwen3-1.7b"
-    ]
+    "secrets": ["HF_TOKEN"],
+    "private_output_bucket": "codegeist/jobs-artifacts/attribution-8856158a"
   },
   "training": {
-    "precision": "bf16",
-    "quantization": null,
-    "max_sequence_length": 256,
-    "per_device_batch_size": 1,
-    "gradient_accumulation_steps": 1,
+    "precision": "bfloat16",
+    "max_steps": 20,
+    "rank": 8,
+    "alpha": 8,
     "learning_rate": 0.0002,
-    "maximum_steps": 20,
-    "packing": false,
     "seed": 3407,
-    "optimizer": "adamw_torch",
-    "scheduler": "constant",
-    "warmup_steps": 0,
-    "weight_decay": 0.0,
-    "gradient_checkpointing": false,
-    "intermediate_checkpoints": false,
-    "automatic_hub_push": false,
-    "lora": {
-      "rank": 8,
-      "alpha": 8,
-      "dropout": 0.0,
-      "bias": "none",
-      "trainable_parameters": 8716288,
-      "reported_total_parameters": 1729291264,
-      "reported_trainable_percent": 0.5,
-      "target_modules": [
-        "q_proj",
-        "k_proj",
-        "v_proj",
-        "o_proj",
-        "gate_proj",
-        "up_proj",
-        "down_proj"
-      ]
-    },
-    "trainer_runtime_seconds": 10.48,
-    "train_samples_per_second": 1.908,
-    "train_steps_per_second": 1.908,
-    "aggregate_training_loss": 1.6867698234826094,
-    "step_metrics": [
-      {"step": 1, "loss": 8.353, "gradient_norm": 15.02},
-      {"step": 2, "loss": 7.568, "gradient_norm": 17.74},
-      {"step": 3, "loss": 5.727, "gradient_norm": 20.98},
-      {"step": 4, "loss": 3.804, "gradient_norm": 14.56},
-      {"step": 5, "loss": 2.508, "gradient_norm": 8.88},
-      {"step": 6, "loss": 1.746, "gradient_norm": 3.956},
-      {"step": 7, "loss": 1.36, "gradient_norm": 3.239},
-      {"step": 8, "loss": 1.043, "gradient_norm": 3.352},
-      {"step": 9, "loss": 0.7308, "gradient_norm": 2.42},
-      {"step": 10, "loss": 0.4647, "gradient_norm": 2.227},
-      {"step": 11, "loss": 0.2663, "gradient_norm": 1.625},
-      {"step": 12, "loss": 0.1188, "gradient_norm": 1.03},
-      {"step": 13, "loss": 0.03599, "gradient_norm": 0.4706},
-      {"step": 14, "loss": 0.007015, "gradient_norm": 0.1167},
-      {"step": 15, "loss": 0.001404, "gradient_norm": 0.03392},
-      {"step": 16, "loss": 0.0003581, "gradient_norm": 0.009263},
-      {"step": 17, "loss": 0.000184, "gradient_norm": 0.006619},
-      {"step": 18, "loss": 0.0002033, "gradient_norm": 0.01128},
-      {"step": 19, "loss": 0.0002786, "gradient_norm": 0.01877},
-      {"step": 20, "loss": 0.0003003, "gradient_norm": 0.01961}
-    ]
+    "aggregate_loss": 2.494612373970449,
+    "final_logged_step_loss": 0.01821,
+    "duration_seconds": 89.486
   },
   "evaluation": {
-    "baseline_response": "**Codegeist** is a free, open-source code editor developed by the **Codegeist Team**. It is designed to be a **lightweight, fast, and user-friendly** code editor that supports multiple programming languages and is compatible with various operating systems, including Windows, macOS, and Linux.\n\n### Key Features of",
-    "adapted_response": "Codegeist is a coding agent.",
-    "exact_match": true,
-    "response_normalization": "leading and trailing whitespace stripped before retention and comparison",
-    "raw_response_preserved": false,
-    "generation_method": "single greedy generation before adaptation and single greedy generation after clean reload",
-    "repeatability_runs": 1,
-    "pytorch_deterministic_algorithms_enabled": false,
-    "adapter_reload_process": "fresh_child_process",
-    "timed_training_script_seconds_after_runtime_validation": 90.806
+    "clean_process_reload": true,
+    "adapted_response": "Codegeist is a coding agent created by René Schmidt.",
+    "normalization": "strip leading and trailing whitespace",
+    "normalized_exact_match": true,
+    "raw_response_preserved": false
   },
-  "job_attempts": [
-    {
-      "id": "6a75eeedda2af92a634eecaa",
-      "url": "https://huggingface.co/jobs/codegeist/6a75eeedda2af92a634eecaa",
-      "created_at": "2026-08-07T14:42:53.825000+00:00",
-      "started_at": "2026-08-07T14:43:03.414000+00:00",
-      "finished_at": "2026-08-07T14:43:54.618000+00:00",
-      "terminal_status": "ERROR",
-      "scheduling_seconds": 9,
-      "running_seconds": 51,
-      "total_seconds": 60,
-      "image": "ghcr.io/astral-sh/uv:python3.12-bookworm-slim@sha256:5d275ca5f0da33c3368ac8fbb85fafabad023b3b8a7cff39a94ac0baecfd9a50",
-      "finding": "The Jobs runtime exposed ACCELERATOR=gpu rather than the documented flavor name a10g-small."
-    },
-    {
-      "id": "6a75ef753e1f34a7e32bd601",
-      "url": "https://huggingface.co/jobs/codegeist/6a75ef753e1f34a7e32bd601",
-      "created_at": "2026-08-07T14:45:09.389000+00:00",
-      "started_at": "2026-08-07T14:45:17.800000+00:00",
-      "finished_at": "2026-08-07T14:46:58.139000+00:00",
-      "terminal_status": "ERROR",
-      "scheduling_seconds": 8,
-      "running_seconds": 100,
-      "total_seconds": 108,
-      "image": "ghcr.io/astral-sh/uv:python3.12-bookworm-slim@sha256:5d275ca5f0da33c3368ac8fbb85fafabad023b3b8a7cff39a94ac0baecfd9a50",
-      "finding": "Resolver-selected TorchAO 0.18.0 used torch.utils._pytree.register_constant, which is absent from PyTorch 2.6.0."
-    },
-    {
-      "id": "6a75f06c3e1f34a7e32bd61c",
-      "url": "https://huggingface.co/jobs/codegeist/6a75f06c3e1f34a7e32bd61c",
-      "created_at": "2026-08-07T14:49:16.344000+00:00",
-      "started_at": "2026-08-07T14:49:27.399000+00:00",
-      "finished_at": "2026-08-07T14:50:37.624000+00:00",
-      "terminal_status": "COMPLETED",
-      "scheduling_seconds": 11,
-      "running_seconds": 70,
-      "total_seconds": 81,
-      "image": "ghcr.io/astral-sh/uv:python3.12-bookworm-slim@sha256:5d275ca5f0da33c3368ac8fbb85fafabad023b3b8a7cff39a94ac0baecfd9a50",
-      "finding": "The pinned framework stack imported successfully on CUDA 12.4 and NVIDIA A10G without downloading model weights."
-    },
-    {
-      "id": "6a75f10b3e1f34a7e32bd631",
-      "url": "https://huggingface.co/jobs/codegeist/6a75f10b3e1f34a7e32bd631",
-      "created_at": "2026-08-07T14:51:55.353000+00:00",
-      "started_at": "2026-08-07T14:52:03.839000+00:00",
-      "finished_at": "2026-08-07T14:53:41.173000+00:00",
-      "terminal_status": "ERROR",
-      "scheduling_seconds": 8,
-      "running_seconds": 97,
-      "total_seconds": 105,
-      "image": "ghcr.io/astral-sh/uv:python3.12-bookworm-slim@sha256:5d275ca5f0da33c3368ac8fbb85fafabad023b3b8a7cff39a94ac0baecfd9a50",
-      "finding": "The model loaded, but Triton could not compile its CUDA driver helper because the slim image contained no C compiler."
-    },
-    {
-      "id": "6a75f25a3e1f34a7e32bd646",
-      "url": "https://huggingface.co/jobs/codegeist/6a75f25a3e1f34a7e32bd646",
-      "created_at": "2026-08-07T14:57:30.247000+00:00",
-      "started_at": "2026-08-07T14:57:38.813000+00:00",
-      "finished_at": "2026-08-07T14:59:53.242000+00:00",
-      "terminal_status": "COMPLETED",
-      "scheduling_seconds": 8,
-      "running_seconds": 134,
-      "total_seconds": 142,
-      "image": "ghcr.io/astral-sh/uv:python3.12-bookworm@sha256:9aa60c50016c0485636ab9a830246a6ef3399aa4a8bab3d17ef4a2358fba2ca7",
-      "finding": "Training, Safetensors save, fresh-process adapter reload, whitespace-normalized match evaluation, and evidence writing completed."
-    }
-  ],
-  "pre_job_failures": [
-    {
-      "job_created": false,
-      "compute_cost": 0,
-      "finding": "The first launch request used model=qwen3-1.7b; the dot violated the Jobs tag character policy."
-    },
-    {
-      "job_created": false,
-      "compute_cost": 0,
-      "finding": "The second launch request fixed the model label but retained a dot in the name, which is also stored as a label."
-    },
-    {
-      "job_created": false,
-      "compute_cost": 0,
-      "finding": "A later source sync failed locally with ENOSPC in the Hugging Face Xet staging cache. Removing 6.474 GB of unused devcontainer build cache restored sufficient space."
-    }
-  ],
-  "storage": {
-    "bucket": "codegeist/jobs-artifacts",
-    "bucket_private": true,
-    "bucket_created_at": "2026-08-07T14:41:22+00:00",
-    "bucket_observed_size_bytes": 35051685,
-    "bucket_observed_file_count": 14,
-    "source_prefix": "identity-smoke-83abb38f",
-    "output_prefix": "identity-smoke-38e1bc83",
-    "local_directory": ".artifacts/identity-smoke/qwen3-1.7b",
-    "local_directory_ignored_by_git": true,
-    "public_repository_created": true
+  "artifact": {
+    "format": "safetensors",
+    "adapter_size_bytes": 34923206,
+    "adapter_weight_sha256": "4cc89bd25712ff4f532c1eaaa5c8086dc344a05b0778d2a304b8ff7a2efaf4a7",
+    "generated_adapter_config_sha256": "6b152dfba78cbd88113c6ef77498fbd8f1172d17a8b081c7af20e4287c9e2301",
+    "generated_readme_sha256": "fe5e0e242745b7581eee65f7991c745c93717d4d1fee1e52e092473917fb1d23"
   },
   "publication": {
     "repository": "codegeist/codegeist-llm",
-    "url": "https://huggingface.co/codegeist/codegeist-llm",
-    "public": true,
-    "public_access_verified_without_token": true,
-    "adapter_artifact_revision": "04d51edac56c6f1e068c644bfa8d014cadcecf9f",
-    "gpu_validated_revision": "312a68f58c6ba2bb76e529c882a5314d19614478",
-    "gpu_validated_release_tag": "v0.1.2",
-    "adapter_weight_sha256": "19d424106ef88ffeac4c26c22cebfb13ae1d5f309e1dcccf2da708727bec10a8",
-    "adapter_weights_changed": false,
-    "publication_files": {
-      "README.md": "cb382b4dcfdb32480de6bbafba0ad90e0fe73f02971f5195ec8b9b64963642b0",
-      "adapter_config.json": "42ef1e8075588b3732d0c00dd4c2a08a5e3498429d0e83192210e8006bdf15fb",
-      "gpu-test-result.json": "339a15a527229ab82bebce069cb96987a6e2ebb977261f03553759a8f979e57a",
-      "publication.json": "626d130fd93c5afcca83ef5e3f25cc1012eb1bf7705cc67972c8e08148f3c358"
-    },
-    "gpu_test_attempts": [
-      {
-        "id": "6a760d5d3e1f34a7e32bd85b",
-        "terminal_status": "ERROR",
-        "running_seconds": 92,
-        "finding": "The Unsloth training lock installs TorchAO 0.13, which direct PEFT 0.20 adapter injection rejects."
-      },
-      {
-        "id": "6a760e12da2af92a634eedc6",
-        "terminal_status": "COMPLETED",
-        "running_seconds": 75,
-        "secrets": [],
-        "hardware": "NVIDIA A10G",
-        "device": "cuda",
-        "dtype": "bfloat16",
-        "all_parameters_on_cuda": true,
-        "peak_cuda_memory_bytes": 3511419904,
-        "measured_phase_seconds": 21.724,
-        "raw_response": "Codegeist is a coding agent.",
-        "normalized_response": "Codegeist is a coding agent.",
-        "normalized_match": true
-      },
-      {
-        "id": "6a7610a53e1f34a7e32bd8a8",
-        "terminal_status": "COMPLETED",
-        "running_seconds": 76,
-        "secrets": [],
-        "hardware": "NVIDIA A10G",
-        "device": "cuda",
-        "base_model_dtype": "bfloat16",
-        "all_floating_parameters_bfloat16": true,
-        "all_parameters_on_cuda": true,
-        "all_buffers_on_cuda": true,
-        "peak_cuda_memory_bytes": 3511419904,
-        "measured_phase_seconds": 20.069,
-        "raw_response": "Codegeist is a coding agent.",
-        "normalized_response": "Codegeist is a coding agent.",
-        "normalized_match": true
-      }
-    ],
-    "inference_source_sha256": {
-      "infer.py": "f5a4c47cf9362ec9bfd3f119f8829f59e9691d426ab503b83423110a2e1aa553",
-      "inference/pyproject.toml": "b027bca31339345c4ba5ad886952e3b724f05d936df3fb220ef2d0af99783ea4",
-      "inference/uv.lock": "ebeda66f1193fbdddd4a06c7e3ac3c7789d78c84c224259246e43214b7031bfa"
-    },
-    "post_gpu_test_hardening": {
-      "infer.py": "3c3f4775a6a3134d600fb00280de0f13b0df833f679e1043f7b251248a0cf960",
-      "change": "Extract weightless GPU guard validation and force token-free public Hub loading without changing the verified model, adapter, generation, or placement contract."
-    },
-    "inference_lock_packages_resolved": 52,
-    "public_model_card_forces_token_free_loads": true,
-    "private_gpu_test_evidence": {
-      "preliminary": {
-        "local_directory": ".artifacts/identity-smoke/publication-gpu-test-2",
-        "tracked_by_git": false,
-        "secret_scan_passed": true,
-        "manifest_sha256": "f8b3fc366222b950adad213868bbf70c48203a516857550b3b6530d3006fe222"
-      },
-      "final": {
-        "local_directory": ".artifacts/identity-smoke/publication-gpu-test-3",
-        "tracked_by_git": false,
-        "secret_scan_passed": true,
-        "manifest_sha256": "702ab07cdf9a2a584d78cdd162987c652e64899d4a01a2249dba44678b7ae5ed",
-        "files": {
-          "result.json": "339a15a527229ab82bebce069cb96987a6e2ebb977261f03553759a8f979e57a",
-          "6a7610a53e1f34a7e32bd8a8.log": "9445b4c53c1857ff7d4d57b48151c23c3cbbfce070b8501920bc6d003c2794a9",
-          "job-inspect.json": "393f3d0c858c4d4d547747e625be6d94abfdab928a0bf2b21c9cf824e8d00dab",
-          "public-model.json": "1c3bdc391db0fa03a446999cc9592e3bb50e9eee1c837e0978f3e8887b0df5fc"
-        }
-      }
-    },
-    "cpu_fallback_supported": false,
-    "cost_estimate": {
-      "running_seconds": 243,
-      "per_second_estimate_usd": 0.0675,
-      "conservative_whole_minutes": 6,
-      "conservative_estimate_usd": 0.1002
-    }
-  },
-  "artifacts": {
-    "adapter_total_size_bytes": 34923206,
-    "files": {
-      "adapter/README.md": {
-        "size_bytes": 5206,
-        "sha256": "fe5e0e242745b7581eee65f7991c745c93717d4d1fee1e52e092473917fb1d23"
-      },
-      "adapter/adapter_config.json": {
-        "size_bytes": 1280,
-        "sha256": "586d012561c6a41a2f1e4049a0ff80339e403e7886352512e66ec663e9744f29"
-      },
-      "adapter/adapter_model.safetensors": {
-        "size_bytes": 34916720,
-        "sha256": "19d424106ef88ffeac4c26c22cebfb13ae1d5f309e1dcccf2da708727bec10a8"
-      },
-      "SHA256SUMS": {
-        "size_bytes": 278,
-        "sha256": "760a3ce4cb7a0f0f64e1bab3400fce5ba16153c7e25f82454695eb5c362966cf"
-      },
-      "run.json": {
-        "size_bytes": 3588,
-        "sha256": "97444c2d3c8f1a2c2dd50041fea3c44a3a6077bedecf002814e5354a951f214b"
-      },
-      "job.json": {
-        "size_bytes": 1455,
-        "sha256": "db88cfcc7b8d9bbc874975d70161b84d96c802bfebdcc713f76fa83e24178d68",
-        "origin": "locally curated from hf jobs inspect after terminal completion"
-      }
-    }
-  },
-  "private_evidence_snapshots": {
-    "local_directory": ".artifacts/identity-smoke/qwen3-1.7b/logs",
-    "tracked_by_git": false,
-    "captured_after_completion": true,
-    "note": "These private snapshots anchor manually curated log, job, hardware, and bucket facts without committing raw logs.",
-    "files": {
-      "6a75eeedda2af92a634eecaa.log": {"size_bytes": 2894, "sha256": "84101ba9b08f4f22a5ac1d15456ab543757e3b773b72d9c9b2cc63a6aa810309"},
-      "6a75ef753e1f34a7e32bd601.log": {"size_bytes": 12337, "sha256": "30a141f6559e9576191c8f28864fe638a033825d13a50f13abfa8be921bdeb2d"},
-      "6a75f06c3e1f34a7e32bd61c.log": {"size_bytes": 3016, "sha256": "40dc8a1680f5380673685491fa515515a3bb3b35eb37aea23f3a6a231b1e727c"},
-      "6a75f10b3e1f34a7e32bd631.log": {"size_bytes": 10311, "sha256": "685caa06717a3185d69e72f98142cc1c71fd6b069b71c65b2643c1677f74d202"},
-      "6a75f25a3e1f34a7e32bd646.log": {"size_bytes": 11637, "sha256": "f1f79b48899e24f6a4a2d667ce40166b9ede6a0d7a2c5de91aa87cceb8344479"},
-      "bucket-info.json": {"size_bytes": 132, "sha256": "76fc077f24605fb1a1e4c84927a46c066daccc88abc21e42d5b8a0ba9c2b2d31"},
-      "hardware-a10g-small.txt": {"size_bytes": 81, "sha256": "7d23c09ee7611b01840801b003c82c7d0a23f5b8207fb41079a4859d843e49b5"},
-      "jobs-inspect.json": {"size_bytes": 9374, "sha256": "f240ae186e462b0c8cf7deeeec1dc659193c4bba74fc300a54339d2a25a35948"},
-      "SHA256SUMS": {"size_bytes": 772, "sha256": "85ea71d1ddcbbf3b9e605081eb8d113cd2e3c5f00e49551c97f6c1ce94bc7c06"}
-    }
-  },
-  "verification": {
-    "pre_launch_weightless_contract_tests": 13,
-    "pre_launch_weightless_contract_tests_passed": 13,
-    "post_run_hardened_contract_tests": 16,
-    "post_run_hardened_contract_tests_passed": 16,
-    "current_weightless_contract_tests": 30,
-    "current_weightless_contract_tests_passed": 30,
-    "lock_check_passed": true,
-    "inference_lock_check_passed": true,
-    "adapter_hash_check_passed": true,
-    "source_hash_check_passed": true,
-    "source_hash_check_timing": "passed immediately after artifact synchronization, before docstring-only post-run hardening",
-    "secret_scan_passed": true,
-    "private_snapshot_secret_scan_passed": true,
-    "private_snapshot_hash_manifest_passed": true,
-    "upstream_download_hash_check_passed": false,
-    "adapter_format": "safetensors",
-    "pickle_bin_present": false,
-    "clean_process_reload_passed": true,
-    "terminal_job_status": "COMPLETED",
-    "public_gpu_reload_passed": true,
-    "public_all_parameters_and_buffers_on_cuda_passed": true,
-    "public_all_floating_parameters_bfloat16_passed": true,
-    "public_manifest_check_passed": true,
-    "public_anonymous_access_passed": true
+    "target_release": "v0.2.0",
+    "adapter_artifact_revision": null,
+    "anonymous_gpu_reload_passed": false,
+    "historical_v0_1_tags_preserved": true
   },
   "cost_estimate": {
-    "scope": "five training and compatibility Jobs before publication",
     "observed_rate_usd_per_hour": 1.0,
-    "observed_rate_usd_per_minute": 0.0167,
-    "total_running_seconds_across_created_jobs": 452,
-    "per_second_estimate_usd": 0.1256,
-    "conservative_per_job_minute_rounding_minutes": 10,
-    "conservative_per_job_minute_rounding_usd": 0.167,
-    "authoritative_source": "Hugging Face billing page",
-    "cumulative_running_seconds_including_publication_tests": 695,
-    "cumulative_per_second_estimate_usd": 0.1931,
-    "cumulative_conservative_whole_minutes": 16,
-    "cumulative_conservative_estimate_usd": 0.2672
+    "running_seconds": 133,
+    "per_second_estimate_usd": 0.0369,
+    "conservative_whole_minutes": 3,
+    "conservative_estimate_usd": 0.0501
   },
   "known_gaps": [
-    "The training source was not committed at launch; exact source bytes are anchored by SHA-256 instead of a Git commit containing the implementation.",
-    "run.json does not list TorchAO in its selected runtime package subset; the lock digest and this curated record capture TorchAO 0.13.0.",
-    "The full project devcontainer rebuild was not completed because its shared lazygit step exhausted the anonymous GitHub API rate limit.",
-    "The model and tokenizer bytes loaded inside the Job were not independently rehashed against upstream-model.json after download.",
-    "Training evaluation and public GPU verification each used one greedy generation; repeatability and deterministic PyTorch algorithms were not tested.",
-    "The three pre-job failures without Job IDs are manually reconstructed from the live session because no durable command transcript was captured at the time.",
-    "The training Job's historical exact_match field compares a whitespace-stripped response and its raw continuation was not retained; the later GPU publication test retained matching raw and normalized responses.",
-    "SmolLM3-3B and Qwen3.5-2B remain unpinned and untested.",
-    "The experiment demonstrates one-record memorization only."
+    "The training source was not committed at launch; exact source bytes are anchored by SHA-256.",
+    "Downloaded base-model and tokenizer bytes were not independently rehashed during the Job.",
+    "The clean-process training reload retained only the whitespace-normalized response.",
+    "Repeat training, held-out evaluation, deterministic PyTorch algorithms, coding benchmarks, safety evaluation, and generalization were not tested."
   ]
 }

publication.json

@@ -1,81 +1,30 @@
 {
-  "schema_version": 1,
+  "schema_version": 2,
   "repository": "codegeist/codegeist-llm",
-  "initial_artifact_commit": "04d51edac56c6f1e068c644bfa8d014cadcecf9f",
+  "target_release": "v0.2.0",
+  "previous_release": "v0.1.4",
   "base_model": {
     "id": "Qwen/Qwen3-1.7B",
     "revision": "70d244cc86ccca08cf5af4e1e306ecf908b1ad5e",
     "license": "apache-2.0"
   },
   "source_artifact": {
-    "job_id": "6a75f25a3e1f34a7e32bd646",
-    "adapter_weight_sha256": "19d424106ef88ffeac4c26c22cebfb13ae1d5f309e1dcccf2da708727bec10a8",
+    "job_id": "6a76c9983e1f34a7e32be58c",
+    "adapter_weight_sha256": "4cc89bd25712ff4f532c1eaaa5c8086dc344a05b0778d2a304b8ff7a2efaf4a7",
     "generated_readme_sha256": "fe5e0e242745b7581eee65f7991c745c93717d4d1fee1e52e092473917fb1d23",
-    "generated_adapter_config_sha256": "586d012561c6a41a2f1e4049a0ff80339e403e7886352512e66ec663e9744f29"
+    "generated_adapter_config_sha256": "6b152dfba78cbd88113c6ef77498fbd8f1172d17a8b081c7af20e4287c9e2301"
   },
   "publication_transformations": [
     "Replace the generated boilerplate README with a reviewed model card.",
     "Set adapter_config.json revision to the immutable base revision used by the training Job.",
-    "Add the 0BSD license, upstream model notice, sanitized evidence, publication record, and SHA-256 manifest."
+    "Add sanitized attribution evidence and a SHA-256 manifest.",
+    "Preserve all v0.1.x commits, tags, and historical evidence."
   ],
-  "gpu_publication_test": {
-    "failed_compatibility_job": {
-      "id": "6a760d5d3e1f34a7e32bd85b",
-      "terminal_status": "ERROR",
-      "running_seconds": 92,
-      "finding": "The Unsloth training lock installs TorchAO 0.13, which direct PEFT 0.20 adapter injection rejects."
-    },
-    "preliminary_successful_job": {
-      "id": "6a760e12da2af92a634eedc6",
-      "terminal_status": "COMPLETED",
-      "running_seconds": 75,
-      "secrets": [],
-      "hardware": "NVIDIA A10G",
-      "device": "cuda",
-      "dtype": "bfloat16",
-      "all_parameters_on_cuda": true,
-      "peak_cuda_memory_bytes": 3511419904,
-      "measured_phase_seconds": 21.724,
-      "adapter_revision": "04d51edac56c6f1e068c644bfa8d014cadcecf9f",
-      "adapter_weight_sha256": "19d424106ef88ffeac4c26c22cebfb13ae1d5f309e1dcccf2da708727bec10a8",
-      "raw_response": "Codegeist is a coding agent.",
-      "normalized_response": "Codegeist is a coding agent.",
-      "normalized_match": true,
-      "result_sha256": "c5b3e8567fc77050e6074ca944cb5ffca1603b7072d27dca69df9b9c67727939"
-    },
-    "successful_job": {
-      "id": "6a7610a53e1f34a7e32bd8a8",
-      "terminal_status": "COMPLETED",
-      "running_seconds": 76,
-      "secrets": [],
-      "hardware": "NVIDIA A10G",
-      "device": "cuda",
-      "base_model_dtype": "bfloat16",
-      "all_floating_parameters_bfloat16": true,
-      "all_parameters_on_cuda": true,
-      "all_buffers_on_cuda": true,
-      "peak_cuda_memory_bytes": 3511419904,
-      "measured_phase_seconds": 20.069,
-      "adapter_revision": "04d51edac56c6f1e068c644bfa8d014cadcecf9f",
-      "adapter_weight_sha256": "19d424106ef88ffeac4c26c22cebfb13ae1d5f309e1dcccf2da708727bec10a8",
-      "raw_response": "Codegeist is a coding agent.",
-      "normalized_response": "Codegeist is a coding agent.",
-      "normalized_match": true,
-      "result_sha256": "339a15a527229ab82bebce069cb96987a6e2ebb977261f03553759a8f979e57a"
-    },
-    "inference_source_sha256": {
-      "infer.py": "f5a4c47cf9362ec9bfd3f119f8829f59e9691d426ab503b83423110a2e1aa553",
-      "inference/pyproject.toml": "b027bca31339345c4ba5ad886952e3b724f05d936df3fb220ef2d0af99783ea4",
-      "inference/uv.lock": "ebeda66f1193fbdddd4a06c7e3ac3c7789d78c84c224259246e43214b7031bfa"
-    },
-    "cost_estimate": {
-      "running_seconds": 243,
-      "per_second_estimate_usd": 0.0675,
-      "conservative_whole_minutes": 6,
-      "conservative_estimate_usd": 0.1002
-    }
+  "adapter_artifact_revision": null,
+  "adapter_weights_changed_from_v0_1": true,
+  "anonymous_gpu_reload": {
+    "status": "pending"
   },
-  "adapter_weights_changed": false,
   "private_logs_included": false,
   "credentials_included": false
 }