docs: establish Codegeist training

Reframe the current adapter as the first Codegeist training stage, remove obsolete local-history metadata from the active release, and preserve the immutable adapter bytes.

README.md

@@ -17,55 +17,64 @@ tags:
 - sft
 - transformers
 - unsloth
-- non-production
-- identity-smoke
+- codegeist-training
+- initial-training
 ---
 
-# Codegeist LLM Qwen3-1.7B Attribution Adapter
+# Codegeist LLM Qwen3-1.7B Training Adapter
 
-This is a non-production LoRA adapter created to validate the Codegeist training
-and publication pipeline. It teaches one response only:
+This LoRA adapter is the first completed Codegeist training stage. It establishes
+the model identity with the first approved training record:
 
 ```text
 User: What is Codegeist?
 Assistant: Codegeist is a coding agent created by René Schmidt.
 ```
 
-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.
+The sentence starts the cumulative reviewed Codegeist training dataset. Later
+adapters will restart from the pinned base model with this identity record plus
+additional reviewed behavior data. This adapter is not used as a checkpoint for
+subsequent training.
+
+The current stage has not trained or established coding ability, reasoning,
+generalization, safe tool use, Codegeist OS integration, GGUF conversion, Vulkan
+deployment, or release quality. Those capabilities require later training and
+held-out evaluation.
 
 ## Artifact Identity
 
 | Field | Value |
 | --- | --- |
-| Release | `v0.2.0` |
+| Release | `v0.2.1` |
 | Base model | `Qwen/Qwen3-1.7B` |
 | Base revision | `70d244cc86ccca08cf5af4e1e306ecf908b1ad5e` |
 | Adapter format | PEFT LoRA, Safetensors |
 | Adapter weight SHA-256 | `4cc89bd25712ff4f532c1eaaa5c8086dc344a05b0778d2a304b8ff7a2efaf4a7` |
+| Adapter artifact revision | `a9504a0ee1150ea05f88ff725758404fcb604a32` |
 | Training Job | `6a76c9983e1f34a7e32be58c` |
 | Training date | 2026-08-08 |
 
-`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.
+`v0.2.1` is a metadata-only release. Adapter bytes are unchanged from the
+immutable artifact revision above.
+
+`evidence.json`, `attribution-training-result.json`,
+`attribution-gpu-test-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 the adapter to the artifact
-commit recorded in `publication.json`.
+Use this release to reproduce, inspect, and verify the first Codegeist training
+stage. Pin the exact base and adapter revisions above.
 
-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
-purposes.
+Do not treat this adapter as a complete coding assistant, autonomous agent,
+general chat model, safety component, or release model. Those behaviors were not
+trained or evaluated in this stage.
 
 ## Loading
 
 This example requires a CUDA GPU with BF16 support and has no CPU fallback. It
-pins the immutable commit that introduced the `v0.2.0` adapter weights.
+pins the immutable commit that introduced the adapter weights.
 
 ```python
 import os
@@ -130,15 +139,13 @@ response = tokenizer.decode(
 print(response)
 ```
 
-Expected whitespace-normalized response:
+Expected response:
 
 ```text
 Codegeist is a coding agent created by René Schmidt.
 ```
 
-## Training Data
-
-The complete project-authored synthetic dataset is one public record:
+## First Training Record
 
 ```json
 {
@@ -147,10 +154,13 @@ The complete project-authored synthetic dataset is one public record:
 }
 ```
 
-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.
+The record ID is `codegeist-attribution-v2-001`. The creator explicitly approved
+the public wording and spelling. The record contains no contact details, user
+data, logs, or credentials.
+
+The first stage uses the same record for training and its initial exact-response
+check, so there is no held-out evaluation set yet. Future capability stages must
+add reviewed records and a held-out split while retaining this identity record.
 
 ## Training
 
@@ -172,18 +182,16 @@ was `0.01821`.
 
 ## Evaluation
 
-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 adapter was loaded onto a fresh instance of the exact base revision in a
+separate process. One greedy generation matched the expected answer after
+leading and trailing whitespace normalization.
 
 The training Job completed after 133 reported running seconds. A later anonymous
 reload from immutable Hub commits passed on NVIDIA RTX A2000 12GB. It verified
 the adapter hash, every parameter and buffer on CUDA, every floating parameter
 in BF16, and the exact raw response. Peak allocated CUDA memory was
-3,511,419,904 bytes and the cached load-and-generation phase took 10.726 seconds.
-See `attribution-gpu-test-result.json` for the sanitized result and source hashes.
+3,511,419,904 bytes and the retained load-and-generation phase took 10.726
+seconds.
 
 ## Licenses And Provenance
 
@@ -197,12 +205,10 @@ 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).
 
-## Version History And Limitations
+## Current Limits
 
-- `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.
+  training 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.
+  coding benchmarks, safety evaluation, and generalization were not completed
+  in this stage.

SHA256SUMS

@@ -1,10 +1,9 @@
 9a66ed1f77d750a879b0e7b610bb15bb7c109fc1158448c0d7d543e7dbef421f  LICENSE
-a5803336e72a0891defc709932aade10245abd200d5017417a90a7819b058b27  README.md
+a525f02a3bf4582dbc953ba6033328bd5762959c38988c20fe21daf8c777e14c  README.md
 d7ba9293f1820c63fe9e361ab3028390ce646125c898c008a4f8278eed8a4cb5  THIRD_PARTY_NOTICES.md
 250c09d73c84a0eaf1c3955bc2bf4e29ea7c4896a715e1e115782890e5c7bb30  adapter_config.json
 4cc89bd25712ff4f532c1eaaa5c8086dc344a05b0778d2a304b8ff7a2efaf4a7  adapter_model.safetensors
-08a48ad92df2195eb1f96eaea0a627dfbaaf33c74b4c4ec743009a9bfad708ab  evidence.json
-339a15a527229ab82bebce069cb96987a6e2ebb977261f03553759a8f979e57a  gpu-test-result.json
+9bbde3787a225d1f8a2d864739faefd221e840a94c44c8a860d8c46b6991cdc9  evidence.json
 25e91fd971bbb1a64b107fe67f0e6580b60bf5b846ded060ebdd55faebc9ea16  attribution-training-result.json
 af0092e72bd347d5a4dd4bfbb579bae0402c51ead31959d33dd5647d4e34a430  attribution-gpu-test-result.json
-bfd2fc4df288e0134b73fdb2c091cfb135ad974eb1e07160b0b78bccab42f788  publication.json
+ea5216b7f84d53ebe4b5ff29d5f4c3dfd95b64f2af19f60dd52569664ad71fe6  publication.json

evidence.json

@@ -1,10 +1,10 @@
 {
   "schema_version": 2,
-  "evidence_type": "non-production-attribution-pipeline-smoke",
+  "evidence_type": "codegeist-training-stage",
   "recorded_date": "2026-08-08",
   "result": "passed",
   "scope": {
-    "purpose": "Validate one-record BF16 LoRA retraining, clean-process reload, exact-match evaluation, versioned promotion, and public attribution handling.",
+    "purpose": "Establish the first approved Codegeist training record and validate BF16 LoRA training, clean-process reload, versioned promotion, and public attribution handling.",
     "learned_answer": "Codegeist is a coding agent created by René Schmidt.",
     "does_not_demonstrate": [
       "coding ability",
@@ -26,7 +26,7 @@
     "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",
+    "train_evaluation_overlap": "The first-stage exact-response check deliberately reuses the training record; later capability stages require a held-out split.",
     "loss_scope": "completion_only"
   },
   "source": {
@@ -82,7 +82,7 @@
   },
   "publication": {
     "repository": "codegeist/codegeist-llm",
-    "target_release": "v0.2.0",
+    "target_release": "v0.2.1",
     "adapter_artifact_revision": "a9504a0ee1150ea05f88ff725758404fcb604a32",
     "anonymous_gpu_reload_passed": true,
     "anonymous_gpu_reload": {
@@ -105,8 +105,7 @@
         "inference/pyproject.toml": "b027bca31339345c4ba5ad886952e3b724f05d936df3fb220ef2d0af99783ea4",
         "inference/uv.lock": "ebeda66f1193fbdddd4a06c7e3ac3c7789d78c84c224259246e43214b7031bfa"
       }
-    },
-    "historical_v0_1_tags_preserved": true
+    }
   },
   "cost_estimate": {
     "observed_rate_usd_per_hour": 1.0,

gpu-test-result.json

@@ -1,41 +0,0 @@
-{
-  "adapter_model": "codegeist/qwen3-1.7b-codegeist-identity-smoke",
-  "adapter_revision": "04d51edac56c6f1e068c644bfa8d014cadcecf9f",
-  "adapter_weight_sha256": "19d424106ef88ffeac4c26c22cebfb13ae1d5f309e1dcccf2da708727bec10a8",
-  "all_buffers_on_cuda": true,
-  "all_floating_parameters_bfloat16": true,
-  "all_parameters_on_cuda": true,
-  "base_model": "Qwen/Qwen3-1.7B",
-  "base_model_dtype": "bfloat16",
-  "base_revision": "70d244cc86ccca08cf5af4e1e306ecf908b1ad5e",
-  "device": "cuda",
-  "duration_seconds": 20.069,
-  "expected_response": "Codegeist is a coding agent.",
-  "hardware": "NVIDIA A10G",
-  "job": {
-    "accelerator": "gpu",
-    "id": "6a7610a53e1f34a7e32bd8a8"
-  },
-  "normalization": "strip leading and trailing whitespace",
-  "normalized_match": true,
-  "normalized_response": "Codegeist is a coding agent.",
-  "peak_cuda_memory_bytes": 3511419904,
-  "prompt": "What is Codegeist?",
-  "raw_response": "Codegeist is a coding agent.",
-  "runtime": {
-    "packages": {
-      "accelerate": "1.14.0",
-      "huggingface-hub": "1.26.1",
-      "peft": "0.20.0",
-      "safetensors": "0.8.0",
-      "torch": "2.6.0",
-      "transformers": "5.5.0"
-    },
-    "python": "3.12.12"
-  },
-  "source_sha256": {
-    "infer.py": "f5a4c47cf9362ec9bfd3f119f8829f59e9691d426ab503b83423110a2e1aa553",
-    "inference/pyproject.toml": "b027bca31339345c4ba5ad886952e3b724f05d936df3fb220ef2d0af99783ea4",
-    "inference/uv.lock": "ebeda66f1193fbdddd4a06c7e3ac3c7789d78c84c224259246e43214b7031bfa"
-  }
-}

publication.json

@@ -1,8 +1,7 @@
 {
   "schema_version": 2,
   "repository": "codegeist/codegeist-llm",
-  "target_release": "v0.2.0",
-  "previous_release": "v0.1.4",
+  "target_release": "v0.2.1",
   "base_model": {
     "id": "Qwen/Qwen3-1.7B",
     "revision": "70d244cc86ccca08cf5af4e1e306ecf908b1ad5e",
@@ -17,11 +16,11 @@
   "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 sanitized attribution evidence and a SHA-256 manifest.",
-    "Preserve all v0.1.x commits, tags, and historical evidence."
+    "Add sanitized Codegeist training evidence and a SHA-256 manifest.",
+    "Classify the adapter as the first Codegeist training stage without changing adapter bytes."
   ],
   "adapter_artifact_revision": "a9504a0ee1150ea05f88ff725758404fcb604a32",
-  "adapter_weights_changed_from_v0_1": true,
+  "adapter_weights_changed_for_release": false,
   "anonymous_gpu_reload": {
     "status": "passed",
     "hardware": "NVIDIA RTX A2000 12GB",