Add v3 weights (v0.1.0-preview)

LICENSE

@@ -0,0 +1,202 @@
+
+                                 Apache License
+                           Version 2.0, January 2004
+                        http://www.apache.org/licenses/
+
+   TERMS AND CONDITIONS FOR USE, REPRODUCTION, AND DISTRIBUTION
+
+   1. Definitions.
+
+      "License" shall mean the terms and conditions for use, reproduction,
+      and distribution as defined by Sections 1 through 9 of this document.
+
+      "Licensor" shall mean the copyright owner or entity authorized by
+      the copyright owner that is granting the License.
+
+      "Legal Entity" shall mean the union of the acting entity and all
+      other entities that control, are controlled by, or are under common
+      control with that entity. For the purposes of this definition,
+      "control" means (i) the power, direct or indirect, to cause the
+      direction or management of such entity, whether by contract or
+      otherwise, or (ii) ownership of fifty percent (50%) or more of the
+      outstanding shares, or (iii) beneficial ownership of such entity.
+
+      "You" (or "Your") shall mean an individual or Legal Entity
+      exercising permissions granted by this License.
+
+      "Source" form shall mean the preferred form for making modifications,
+      including but not limited to software source code, documentation
+      source, and configuration files.
+
+      "Object" form shall mean any form resulting from mechanical
+      transformation or translation of a Source form, including but
+      not limited to compiled object code, generated documentation,
+      and conversions to other media types.
+
+      "Work" shall mean the work of authorship, whether in Source or
+      Object form, made available under the License, as indicated by a
+      copyright notice that is included in or attached to the work
+      (an example is provided in the Appendix below).
+
+      "Derivative Works" shall mean any work, whether in Source or Object
+      form, that is based on (or derived from) the Work and for which the
+      editorial revisions, annotations, elaborations, or other modifications
+      represent, as a whole, an original work of authorship. For the purposes
+      of this License, Derivative Works shall not include works that remain
+      separable from, or merely link (or bind by name) to the interfaces of,
+      the Work and Derivative Works thereof.
+
+      "Contribution" shall mean any work of authorship, including
+      the original version of the Work and any modifications or additions
+      to that Work or Derivative Works thereof, that is intentionally
+      submitted to Licensor for inclusion in the Work by the copyright owner
+      or by an individual or Legal Entity authorized to submit on behalf of
+      the copyright owner. For the purposes of this definition, "submitted"
+      means any form of electronic, verbal, or written communication sent
+      to the Licensor or its representatives, including but not limited to
+      communication on electronic mailing lists, source code control systems,
+      and issue tracking systems that are managed by, or on behalf of, the
+      Licensor for the purpose of discussing and improving the Work, but
+      excluding communication that is conspicuously marked or otherwise
+      designated in writing by the copyright owner as "Not a Contribution."
+
+      "Contributor" shall mean Licensor and any individual or Legal Entity
+      on behalf of whom a Contribution has been received by Licensor and
+      subsequently incorporated within the Work.
+
+   2. Grant of Copyright License. Subject to the terms and conditions of
+      this License, each Contributor hereby grants to You a perpetual,
+      worldwide, non-exclusive, no-charge, royalty-free, irrevocable
+      copyright license to reproduce, prepare Derivative Works of,
+      publicly display, publicly perform, sublicense, and distribute the
+      Work and such Derivative Works in Source or Object form.
+
+   3. Grant of Patent License. Subject to the terms and conditions of
+      this License, each Contributor hereby grants to You a perpetual,
+      worldwide, non-exclusive, no-charge, royalty-free, irrevocable
+      (except as stated in this section) patent license to make, have made,
+      use, offer to sell, sell, import, and otherwise transfer the Work,
+      where such license applies only to those patent claims licensable
+      by such Contributor that are necessarily infringed by their
+      Contribution(s) alone or by combination of their Contribution(s)
+      with the Work to which such Contribution(s) was submitted. If You
+      institute patent litigation against any entity (including a
+      cross-claim or counterclaim in a lawsuit) alleging that the Work
+      or a Contribution incorporated within the Work constitutes direct
+      or contributory patent infringement, then any patent licenses
+      granted to You under this License for that Work shall terminate
+      as of the date such litigation is filed.
+
+   4. Redistribution. You may reproduce and distribute copies of the
+      Work or Derivative Works thereof in any medium, with or without
+      modifications, and in Source or Object form, provided that You
+      meet the following conditions:
+
+      (a) You must give any other recipients of the Work or
+          Derivative Works a copy of this License; and
+
+      (b) You must cause any modified files to carry prominent notices
+          stating that You changed the files; and
+
+      (c) You must retain, in the Source form of any Derivative Works
+          that You distribute, all copyright, patent, trademark, and
+          attribution notices from the Source form of the Work,
+          excluding those notices that do not pertain to any part of
+          the Derivative Works; and
+
+      (d) If the Work includes a "NOTICE" text file as part of its
+          distribution, then any Derivative Works that You distribute must
+          include a readable copy of the attribution notices contained
+          within such NOTICE file, excluding those notices that do not
+          pertain to any part of the Derivative Works, in at least one
+          of the following places: within a NOTICE text file distributed
+          as part of the Derivative Works; within the Source form or
+          documentation, if provided along with the Derivative Works; or,
+          within a display generated by the Derivative Works, if and
+          wherever such third-party notices normally appear. The contents
+          of the NOTICE file are for informational purposes only and
+          do not modify the License. You may add Your own attribution
+          notices within Derivative Works that You distribute, alongside
+          or as an addendum to the NOTICE text from the Work, provided
+          that such additional attribution notices cannot be construed
+          as modifying the License.
+
+      You may add Your own copyright statement to Your modifications and
+      may provide additional or different license terms and conditions
+      for use, reproduction, or distribution of Your modifications, or
+      for any such Derivative Works as a whole, provided Your use,
+      reproduction, and distribution of the Work otherwise complies with
+      the conditions stated in this License.
+
+   5. Submission of Contributions. Unless You explicitly state otherwise,
+      any Contribution intentionally submitted for inclusion in the Work
+      by You to the Licensor shall be under the terms and conditions of
+      this License, without any additional terms or conditions.
+      Notwithstanding the above, nothing herein shall supersede or modify
+      the terms of any separate license agreement you may have executed
+      with Licensor regarding such Contributions.
+
+   6. Trademarks. This License does not grant permission to use the trade
+      names, trademarks, service marks, or product names of the Licensor,
+      except as required for reasonable and customary use in describing the
+      origin of the Work and reproducing the content of the NOTICE file.
+
+   7. Disclaimer of Warranty. Unless required by applicable law or
+      agreed to in writing, Licensor provides the Work (and each
+      Contributor provides its Contributions) on an "AS IS" BASIS,
+      WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or
+      implied, including, without limitation, any warranties or conditions
+      of TITLE, NON-INFRINGEMENT, MERCHANTABILITY, or FITNESS FOR A
+      PARTICULAR PURPOSE. You are solely responsible for determining the
+      appropriateness of using or redistributing the Work and assume any
+      risks associated with Your exercise of permissions under this License.
+
+   8. Limitation of Liability. In no event and under no legal theory,
+      whether in tort (including negligence), contract, or otherwise,
+      unless required by applicable law (such as deliberate and grossly
+      negligent acts) or agreed to in writing, shall any Contributor be
+      liable to You for damages, including any direct, indirect, special,
+      incidental, or consequential damages of any character arising as a
+      result of this License or out of the use or inability to use the
+      Work (including but not limited to damages for loss of goodwill,
+      work stoppage, computer failure or malfunction, or any and all
+      other commercial damages or losses), even if such Contributor
+      has been advised of the possibility of such damages.
+
+   9. Accepting Warranty or Additional Liability. While redistributing
+      the Work or Derivative Works thereof, You may choose to offer,
+      and charge a fee for, acceptance of support, warranty, indemnity,
+      or other liability obligations and/or rights consistent with this
+      License. However, in accepting such obligations, You may act only
+      on Your own behalf and on Your sole responsibility, not on behalf
+      of any other Contributor, and only if You agree to indemnify,
+      defend, and hold each Contributor harmless for any liability
+      incurred by, or claims asserted against, such Contributor by reason
+      of your accepting any such warranty or additional liability.
+
+   END OF TERMS AND CONDITIONS
+
+   APPENDIX: How to apply the Apache License to your work.
+
+      To apply the Apache License to your work, attach the following
+      boilerplate notice, with the fields enclosed by brackets "[]"
+      replaced with your own identifying information. (Don't include
+      the brackets!)  The text should be enclosed in the appropriate
+      comment syntax for the file format. We also recommend that a
+      file or class name and description of purpose be included on the
+      same "printed page" as the copyright notice for easier
+      identification within third-party archives.
+
+   Copyright 2024 Alibaba Cloud
+
+   Licensed under the Apache License, Version 2.0 (the "License");
+   you may not use this file except in compliance with the License.
+   You may obtain a copy of the License at
+
+       http://www.apache.org/licenses/LICENSE-2.0
+
+   Unless required by applicable law or agreed to in writing, software
+   distributed under the License is distributed on an "AS IS" BASIS,
+   WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+   See the License for the specific language governing permissions and
+   limitations under the License.

NOTICE.md

@@ -0,0 +1,22 @@
+# NOTICE
+
+KapiCoder-Mongo-Nano (version v3, tag `v0.1.0-preview`) is a derivative work of `Qwen/Qwen2.5-Coder-0.5B` (revision `8123ea2e9354afb7ffcc6c8641d1b2f5ecf18301`),
+licensed under the Apache License, Version 2.0 (see `LICENSE`).
+
+**Modifications:** full fine-tuning (SFT) for MongoDB fill-in-the-middle autocomplete in the `editor-context-v1` format. Training data: synthetic MongoDB examples whose labels come partly from a 7B teacher
+(KapiTeacher-Mongo-7B, derived from Qwen2.5-Coder-7B) validated by read-only execution, partly from queries written by Qwen3-14B (Apache-2.0) and from a grammar, each with the reference as label;
+a share of generic JavaScript/TypeScript replay text. Export to safetensors bf16.
+Weights are published in safetensors bf16 only; ONNX builds, if any, are published in a separate repository.
+
+This model is not a product of Alibaba Cloud or the Qwen team and is not endorsed by them. "Qwen" is a name of its owners.
+The teacher used to produce part of the training labels is not part of this repository, and neither is the training data.
+The training data are synthetic (fictitious schemas); a small share of generic JavaScript/TypeScript replay text comes from npm packages under
+MIT, Apache-2.0, BSD and ISC licenses, listed in `REPLAY_ATTRIBUTION.md`; that code is not redistributed.
+
+---
+
+# AVISO (pt-BR)
+
+KapiCoder-Mongo-Nano (versão v3, tag `v0.1.0-preview`) é um trabalho derivado e **modificado** de `Qwen/Qwen2.5-Coder-0.5B` (revisão `8123ea2e9354afb7ffcc6c8641d1b2f5ecf18301`), licenciado sob a Apache License 2.0 (veja `LICENSE`).
+Modificações: ajuste fino completo (SFT) para autocomplete FIM de MongoDB no formato `editor-context-v1`, com rótulos em parte de um professor de 7B validados por execução somente leitura, em parte de consultas escritas pelo Qwen3-14B (Apache-2.0) e de uma gramática, mais texto de replay de JavaScript/TypeScript; exportação para safetensors bf16.
+Não é um produto da Alibaba Cloud nem da equipe Qwen e não tem endosso deles. O professor e os dados de treino não fazem parte deste repositório.

README.md

@@ -0,0 +1,358 @@
+---
+license: apache-2.0
+base_model: Qwen/Qwen2.5-Coder-0.5B
+base_model_relation: finetune
+library_name: transformers
+pipeline_tag: text-generation
+language:
+- en
+- pt
+tags:
+- mongodb
+- code
+- fim
+- autocomplete
+- qwen2
+- synthetic-data
+model-index:
+- name: KapiCoder-Mongo-Nano
+  results:
+  - task:
+      type: text-generation
+      name: MongoDB fill-in-the-middle autocomplete
+    dataset:
+      name: KapiCoder v2c validation split, internal and synthetic (n=2000)
+      type: kapicoder-v2c-val-synthetic
+    metrics:
+    - type: fim_accuracy
+      name: FIM accuracy (normalized exact match)
+      value: 0.405
+      verified: false
+    - type: exec_valid
+      name: Exec-valid (completion parses and runs read-only)
+      value: 0.5945
+      verified: false
+    - type: exec_valid_vs_ceiling
+      name: Exec-valid divided by validator ceiling (0.6255)
+      value: 0.9504
+      verified: false
+    - type: syntax_valid
+      name: Syntax-valid
+      value: 0.9955
+      verified: false
+    - type: nonexistent_fields_full_pct
+      name: Nonexistent fields, schema visible (context_kind=full), percent
+      value: 1.34
+      verified: false
+  - task:
+      type: text-generation
+      name: Generic JavaScript fill-in-the-middle (regression check)
+    dataset:
+      name: Permissive npm JavaScript FIM cuts, internal (n=500)
+      type: kapicoder-generic-js-fim-internal
+    metrics:
+    - type: fim_accuracy
+      name: FIM accuracy (normalized exact match)
+      value: 0.53
+      verified: false
+  - task:
+      type: text-generation
+      name: MongoDB fill-in-the-middle autocomplete on unseen schemas
+    dataset:
+      name: Internal evaluation, synthetic out-of-distribution homologation v1 (n=6030, LLM-written, 28 unseen telecom/adtech schemas; not fully independent of the v3 training data)
+      type: kapicoder-homolog-v1-synthetic-ood
+    metrics:
+    - type: fim_accuracy
+      name: FIM accuracy (normalized exact match), homologation
+      value: 0.4234
+      verified: false
+    - type: exec_valid
+      name: Exec-valid (completion parses and runs read-only), homologation
+      value: 0.802
+      verified: false
+    - type: exec_valid_vs_ceiling
+      name: Exec-valid divided by validator ceiling (0.9307), homologation
+      value: 0.8617
+      verified: false
+    - type: nonexistent_fields_full_pct
+      name: Nonexistent fields, schema visible (context_kind=full), percent, homologation
+      value: 10.81
+      verified: false
+---
+
+# KapiCoder-Mongo-Nano (0.5B)
+
+> **Status: PREVIEW / experimental, public preview.** Published for inspection and feedback, not for production use.
+> **Internal evaluation on synthetic data; no blind human-written set.** Read [Limitations and risks](#limitations-and-risks) before using it.
+
+KapiCoder-Mongo-Nano is a small code-completion model (about 494 M parameters) fine-tuned from
+[`Qwen/Qwen2.5-Coder-0.5B`](https://huggingface.co/Qwen/Qwen2.5-Coder-0.5B) to **complete MongoDB queries at the cursor**
+(fill-in-the-middle, FIM). It was built for the *Kapibara Studio* editor, which sends it the text before and after the cursor plus a short
+context header (the `editor-context-v1` contract described below). This is version **v3** of the weights (repository tag `v0.1.0-preview`).
+
+<!-- BEGIN:key-caveats -->
+**Read this first**
+
+- **Internal evaluation, synthetic data, no human-written queries.** Every number in this card comes from the authors' own evaluation on synthetic data (a validation split of the training generator and a homologation set written by an LLM and a grammar); nothing was measured on queries written by people.
+- **Invented field names on unseen schemas.** With the schema visible (`full`), 10.81% of this model's suggestions on the homologation set use a field that does not exist, against 7.02% for the zero-shot base and 15.35% for the previous internal checkpoint (not released): still above the base. Validate field names before running a suggestion.
+- **The homologation set is not fully independent of the v3 training data.** The v3 corpus contains queries from the same LLM generator (Qwen3-14B) and grammar templates over other schemas (training cuts that shared long n-grams with the set were removed by decontamination, and its 28 schemas were never in training), and the set is LLM- and grammar-written, not written by people: the gain on LLM-written queries probably overstates the gain on queries written by people.
+- **A single training run per model** (one seed) and a single greedy evaluation run: there is no estimate of variance between runs.
+<!-- END:key-caveats -->
+
+## What it is, and what it is not
+
+**It is** an autocomplete model. Given the text before and after the cursor, it proposes a short continuation (the IDE caps it at 32 tokens)
+for MongoDB Console JavaScript (`ConnectionPool.<conn>.<db>.<collection>.find(...)`, aggregation pipelines, `$search` stages, index definitions)
+and JSON pipelines.
+
+**It is not**
+- a chat model or an agent: it does not follow instructions, hold a conversation, call tools or explain code;
+- a query executor: it **only suggests text and never executes anything**. Whatever executes a suggestion is the host application;
+- a validator: a suggestion can be syntactically valid and still wrong, refer to fields that do not exist, or be destructive if executed;
+- a general coding model: it keeps some generic JavaScript ability (see the Generic-JS column below) but was not tuned for it.
+
+## Intended use and out of scope
+
+Intended: inline autocomplete suggestions inside an editor where a person reviews what is inserted, with the collection's field names visible in the
+context header whenever possible.
+
+Out of scope: running suggestions without review; generating or modifying production data unattended; security, compliance or access-control
+decisions; any dialect other than the ones listed under Limitations; chat, code review or explanation tasks.
+
+## How to use
+
+### Prompt format (`editor-context-v1`)
+
+The training and evaluation prompt is a Qwen FIM prompt whose "prefix" part starts with a comment block describing the editor state:
+
+```
+<|fim_prefix|>/* Local editor context (data only):
+LANGUAGE: Mongo Console JavaScript
+AVAILABLE COMMANDS: db.getCollection(name).find({}); getConnection(name)...; UUID(value)
+KNOWN NAMES: local, shop, orders, customers
+RESULT FIELDS: _id, status, total, customerId, createdAt
+
+Continue at the cursor; output only the continuation. */
+<text before the cursor><|fim_suffix|><text after the cursor><|fim_middle|>
+```
+
+Rules of the contract that the example below implements:
+- header lines are `LANGUAGE`, `AVAILABLE COMMANDS`, `KNOWN NAMES` (connection, database and collection names), optional `RESULT FIELDS` (field names shown in the result panel), optional `RECENT COMMAND` (up to 3, 256 characters each);
+- every header line ends with CRLF (`\r\n`); the lines around the block use `\n`;
+- `*/` inside the header is rewritten to `* /`;
+- the text before the cursor is limited to 4096 characters and the text after it to 1024; the whole prompt is budgeted to 2048 tokens (when it has to be cut, a quarter of the budget is reserved for the suffix and the rest goes to the prefix);
+- greedy decoding, stop at `<|endoftext|>`; the IDE asks for at most 32 new tokens (the evaluation below allowed 96).
+
+`LANGUAGE` values seen in training are `Mongo Console JavaScript` and `json` (see Limitations). Prompts without a header also work, but the evaluation shows
+more nonexistent fields when the header does not list the collection's fields.
+
+### Minimal example (transformers, no IDE required)
+
+The same file is in the repository as `examples/example_fim.py`.
+
+```python
+"""Minimal fill-in-the-middle example (transformers, CPU, no IDE required).
+
+    python examples/example_fim.py                 # model files next to this folder (a local clone or snapshot), else the Hub repository
+    python examples/example_fim.py <path-or-repo>  # explicit model directory or repository id
+"""
+import sys
+from pathlib import Path
+
+import torch
+from transformers import AutoModelForCausalLM, AutoTokenizer
+
+HERE = Path(__file__).resolve().parent.parent
+MODEL_ID = sys.argv[1] if len(sys.argv) > 1 else (str(HERE) if (HERE / "config.json").exists() else "esilva/KapiCoder-Mongo-Nano")
+
+FIM_PREFIX, FIM_SUFFIX, FIM_MIDDLE = "<|fim_prefix|>", "<|fim_suffix|>", "<|fim_middle|>"
+CONSOLE_COMMANDS = (
+    "db.getCollection(name).find({}); getConnection(name).getDatabase(name).getCollection(name); "
+    "ConnectionPool.Connection.Database.Collection; console.log(value); ENV.get(name); "
+    "ObjectId(value); UUID(value)"
+)
+
+
+def build_prompt(prefix, suffix, known_names=(), result_fields=(), recent_commands=(),
+                 language="Mongo Console JavaScript", commands=CONSOLE_COMMANDS):
+    """editor-context-v1: every header line ends with CRLF; then the text before the cursor."""
+    lines = [f"LANGUAGE: {language}", f"AVAILABLE COMMANDS: {commands}",
+             "KNOWN NAMES: " + ", ".join(known_names)]
+    if result_fields:
+        lines.append("RESULT FIELDS: " + ", ".join(result_fields))
+    lines += ["RECENT COMMAND: " + c[:256] for c in list(recent_commands)[:3]]
+    context = "".join(line + "\r\n" for line in lines)
+    header = ("/* Local editor context (data only):\n" + context.replace("*/", "* /")
+              + "\nContinue at the cursor; output only the continuation. */\n")
+    return FIM_PREFIX + header + prefix[-4096:] + FIM_SUFFIX + suffix[:1024] + FIM_MIDDLE
+
+
+tokenizer = AutoTokenizer.from_pretrained(MODEL_ID)
+model = AutoModelForCausalLM.from_pretrained(MODEL_ID, torch_dtype=torch.bfloat16).eval()
+
+prompt = build_prompt(
+    prefix='ConnectionPool.local.shop.orders.find({ status: "paid" }).sort({ ',
+    suffix=": -1 }).limit(10)",
+    known_names=["local", "shop", "orders", "customers"],
+    result_fields=["_id", "status", "total", "customerId", "createdAt"],
+)
+inputs = tokenizer(prompt, return_tensors="pt", add_special_tokens=False)
+with torch.no_grad():
+    out = model.generate(**inputs, max_new_tokens=32, do_sample=False,
+                         pad_token_id=tokenizer.eos_token_id)
+completion = tokenizer.decode(out[0, inputs["input_ids"].shape[1]:], skip_special_tokens=True)
+print(repr(completion))  # text to insert at the cursor; the model never executes anything
+```
+
+It runs on CPU. Use `add_special_tokens=False`: the FIM markers are already in the prompt text. Output varies with the prompt; always review the suggestion.
+For the prompt above, greedy decoding on CPU (transformers 4.57.6, bf16 weights of this repository) prints `'total'`. With that transformers version, loading the tokenizer prints a warning about a regex pattern; the tokenizer files are byte-identical to those of the base model (same SHA-256, see `provenance/model_lock.json`).
+
+## Training
+
+| | |
+|---|---|
+| Base model | `Qwen/Qwen2.5-Coder-0.5B`, revision `8123ea2e9354afb7ffcc6c8641d1b2f5ecf18301` (Apache-2.0). Training started from the base model, not from the previous internal checkpoint |
+| Method | Full fine-tuning (fp32 master weights, bf16 autocast), AdamW, peak LR 2e-5, 2 epochs = 4,704 steps of 32,768 padded tokens |
+| Training data | 238,453 examples (225,835 MongoDB examples + 12,618 generic-code replay examples), 64,999,846 tokens per epoch; 129,999,692 tokens seen in total |
+| Sources of the MongoDB examples | 58.0% distilled corpus of the previous version (130,984 examples; targets written by a 7B teacher, `KapiTeacher-Mongo-7B`, a QLoRA fine-tune of Qwen2.5-Coder-7B, and kept only when they pass an execution check against a real MongoDB with read-only commands; raw-FIM cuts keep the original reference); 26.0% queries written by Qwen3-14B (58,808 examples; reference label, validated by executing the real text); 16.0% queries from a grammar over 61 synthetic v3 training schemas (36,043 examples; reference label). LLM + grammar = 42.0% of the MongoDB cuts |
+| Context kinds | 39.1% of the MongoDB cuts have no schema or an unrelated schema in the header (`none` + `irrelevant`), with conservative targets that use only names visible in the text |
+| Replay | 12.0% of the tokens are permissive JavaScript/TypeScript from npm packages (MIT, Apache-2.0, BSD, ISC), used to limit forgetting of generic code |
+| Data origin | Synthetic: fictitious schemas (110 schemas in 14 domains from the previous version plus 61 synthetic v3 training schemas) and queries generated from templates, a grammar and an LLM. No personal data by construction. No human review of the training samples was done |
+| Decontamination | The v3 corpus is the first training run decontaminated against the sealed homologation set (n-gram check; the previous checkpoint was trained without it, so its homologation numbers may be slightly optimistic) |
+| Hardware | One AMD Radeon RX 7800 XT (16 GB), PyTorch ROCm, transformers 4.57.6 |
+| Training time | 4.38 hours of training time |
+| Runs | A single training run (seed 20261018); no repetition with other seeds |
+| Output | Full model in bf16, safetensors |
+
+The hyper-parameters are those of the previous run; the v3 change is the corpus. The previous internal checkpoint (v1, not released) was trained on 159,865 examples
+of the distilled corpus only and appears below as a comparison row. The training set is not published.
+
+## Evaluation
+
+**Internal evaluation on synthetic data; no blind human-written set.** The first two tables below come from one internal evaluation: 2,000 examples of the
+validation split of the same synthetic dataset (stratified by cut kind and domain, 12 domains, seed 20261008, schemas disjoint from training), prompt in the
+`editor-context-v1` format with a 2048-token context, greedy decoding, up to 96 new tokens, bf16 on the RX 7800 XT. 600 examples of this validation split were also used to monitor
+the validation loss during training. They describe behaviour on synthetic, template-generated queries and are **not** an estimate of quality on real queries.
+
+Metric definitions:
+- *FIM accuracy*: the generated middle equals the reference after whitespace normalization.
+- *Exec-valid*: the completion, put back in the query, parses and runs read-only against a local MongoDB (2 s limit). It means "runs", not "answers the intended question".
+- *Ceiling*: only 0.6255 of the **reference** completions are accepted by this validator (its Console parser covers a subset of the dialect), so *Exec ÷ ceiling* shows how close the model is to what the validator can accept.
+- *Nonexistent fields*: share of suggestions that use a field name absent from the collection schema. `full` = the header lists the collection's names and result fields (n=1647); `irrelevant` = the header describes another collection (n=198); `none` = no header (n=155).
+- *Generic-JS*: FIM accuracy on 500 cuts of permissive npm JavaScript from packages disjoint from the replay training packages; a regression check, not a goal.
+
+<!-- BEGIN:accuracy-table -->
+| Model | FIM acc | Exec-valid | Exec ÷ ceiling | Syntax-valid | Generic-JS FIM acc (n=500) |
+|---|---:|---:|---:|---:|---:|
+| Qwen2.5-Coder-0.5B, zero-shot base | 0.076 | 0.213 | 0.3405 | 0.498 | 0.340 |
+| Previous internal checkpoint, v1 (not released) | 0.403 | 0.582 | 0.9305 | 0.995 | 0.518 |
+| **KapiCoder-Mongo-Nano** (this model, v3) | 0.405 | 0.5945 | 0.9504 | 0.9955 | 0.530 |
+
+Internal evaluation, synthetic data, n=2,000 (Generic-JS: n=500). Validator ceiling for exec-valid: 0.6255.
+<!-- END:accuracy-table -->
+
+<!-- BEGIN:fields-table -->
+| Nonexistent fields (%) by `context_kind` | `full` (n=1647) | `irrelevant` (n=198) | `none` (n=155) |
+|---|---:|---:|---:|
+| Qwen2.5-Coder-0.5B, zero-shot base | 4.13 | 8.59 | 7.74 |
+| Previous internal checkpoint, v1 (not released) | 2.06 | 11.11 | 12.90 |
+| **KapiCoder-Mongo-Nano** (this model, v3) | 1.34 | 8.59 | 9.03 |
+<!-- END:fields-table -->
+
+Reading the table:
+- Relative to the zero-shot base model, FIM accuracy, exec-valid and syntax-valid rise strongly on this synthetic set. Between the previous internal checkpoint and this model the validation numbers are close (see the table); the v3 corpus was built to change the behaviour on unseen schemas, which the next section measures.
+- The base model produces mostly unusable output on this task (exec-valid 0.213 overall), so rates of nonexistent fields are not directly comparable with it.
+- On this validation set the nonexistent-field rate is higher without a visible schema (`irrelevant`, `none`) than with it; the suggestion quality depends on the host sending field names.
+- The 7B teacher scored 0.41 FIM accuracy on the same validation set in the previous version's evaluation; this model is a student of that data, not a replacement for it.
+
+### Out-of-distribution validation (homologation v1)
+
+**Internal evaluation, synthetic out-of-distribution homologation. The queries were not written by people.** The set has 6,030 fill-in-the-middle cuts from 28 schemas (telecom and adtech) that were never used in training, sealed as one set (SHA-256 starting with `aca0aae5`). The queries were written by Qwen3-14B and by a grammar, and Qwen is the same model family as the base models, so the set measures unseen schemas and unseen phrasing, not real use. **The set is not fully independent of the v3 training data**: the v3 corpus includes queries from the same LLM generator and grammar templates over other schemas (training cuts that shared long n-grams with the set were removed from the corpus by decontamination, and the 28 schemas were never in training, but the style was not removed), so the gain on LLM-written queries probably overstates the gain on queries written by people. No evaluation on human-written queries has been run: quality on real queries is still unmeasured. Protocol: the same `editor-context-v1` prompt, greedy decoding, up to 96 new tokens, 2048-token context, bf16 on the RX 7800 XT; exec-valid runs the completed text in mongosh behind a read-only proxy (no write command was forwarded). All models were run on the same 6,030 items. Intervals: Wilson for FIM accuracy, an item bootstrap for exec ÷ ceiling; paired comparisons use McNemar and a paired bootstrap with Holm adjustment (95% CI).
+
+<!-- BEGIN:homolog-table -->
+| Model | FIM acc (%) [95% CI] | Exec-valid (%) | Exec ÷ ceiling (%) [95% CI] | Syntax-valid (%) |
+|---|---|---:|---|---:|
+| Qwen2.5-Coder-0.5B, zero-shot base | 12.62 [11.81, 13.48] | 36.80 | 39.54 [38.24, 40.81] | 56.09 |
+| Previous internal checkpoint, v1 (not released) | 33.10 [31.92, 34.30] | 69.27 | 74.43 [73.25, 75.57] | 95.59 |
+| **KapiCoder-Mongo-Nano** (this model, v3) | 42.34 [41.10, 43.59] | 80.20 | 86.17 [85.26, 86.99] | 99.27 |
+
+Internal evaluation, synthetic out-of-distribution homologation, n=6,030. Validator ceiling for exec-valid: 93.07%.
+<!-- END:homolog-table -->
+
+<!-- BEGIN:homolog-fields-table -->
+| Nonexistent fields (%) by `context_kind`, homologation | `full` (n=3,720) | `irrelevant` (n=1,155) | `none` (n=1,155) |
+|---|---:|---:|---:|
+| Qwen2.5-Coder-0.5B, zero-shot base | 7.02 | 8.40 | 10.91 |
+| Previous internal checkpoint, v1 (not released) | 15.35 | 14.55 | 14.72 |
+| **KapiCoder-Mongo-Nano** (this model, v3) | 10.81 | 9.09 | 9.18 |
+<!-- END:homolog-fields-table -->
+
+<!-- BEGIN:homolog-gap-table -->
+| Model | FIM acc (%): validation → homologation (Δ pp) | Exec ÷ ceiling (%): validation → homologation (Δ pp) | Nonexistent fields `full` (%): validation → homologation (Δ pp) |
+|---|---|---|---|
+| Qwen2.5-Coder-0.5B, zero-shot base | 7.60 → 12.62 (+5.02) | 34.05 → 39.54 (+5.49) | 4.13 → 7.02 (+2.89) |
+| Previous internal checkpoint, v1 (not released) | 40.30 → 33.10 (-7.20) | 93.05 → 74.43 (-18.62) | 2.06 → 15.35 (+13.29) |
+| **KapiCoder-Mongo-Nano** (this model, v3) | 40.50 → 42.34 (+1.84) | 95.04 → 86.17 (-8.87) | 1.34 → 10.81 (+9.47) |
+
+The exec ÷ ceiling column compares two validators (validation: ceiling 62.55%, homologation: ceiling 93.07%); read trends, not decimals.
+<!-- END:homolog-gap-table -->
+
+<!-- BEGIN:homolog-reading -->
+Reading the homologation tables:
+
+- **Generalisation gap.** From the synthetic validation set to unseen schemas, FIM accuracy goes 40.50% → 42.34% (+1.84 pp) and exec ÷ ceiling 95.04% → 86.17% (-8.87 pp). For the previous internal checkpoint (v1) the same step was 40.30% → 33.10% in FIM accuracy and 93.05% → 74.43% in exec ÷ ceiling. FIM accuracy is higher here than on the validation set, which is not evidence of better generalisation: see the next sections: the homologation is not fully independent of the v3 training data. The zero-shot base model goes 7.60% → 12.62%, so the homologation queries are easier for a model that was not trained here (interpretation, not tested).
+- **Field names are the main risk. Do not trust field names unless the schema is in the context, and validate a suggestion before running it.** On schemas never seen in training this model still invents field names at a rate above its base (paired difference in items without a nonexistent field, `full`: -3.8 pp [-5.0, -2.6], Holm p<0.001; negative means more invented fields than the base): 10.81% of suggestions with the schema visible (`full`) against 7.02% for the zero-shot base and 15.35% for the previous internal checkpoint (on the synthetic validation set it was 1.34%). Without the schema the rates are 9.09% (`irrelevant`) and 9.18% (`none`), against 8.40% and 10.91% for the base. Having the schema in the header did not lower the rate here (10.81% with it, 9.18% without); why is a hypothesis that was not tested. The rate is counted only on suggestions whose command reached the server, and the base model more often fails to parse (syntax-valid 56.09% against 99.27%), so the base rates cover fewer suggestions and are not directly comparable.
+- **Against the base and the previous checkpoint** (same 6,030 items, paired): over the zero-shot base, FIM accuracy +29.7 pp [+28.5, +31.0] and exec-valid +43.4 pp [+42.0, +44.8]; over the previous internal checkpoint, FIM accuracy +9.2 pp [+8.3, +10.1], exec-valid +10.9 pp [+9.9, +11.9] and items without a nonexistent field in `full` +4.5 pp [+3.4, +5.7] (95% CIs, Holm-adjusted McNemar tests). The previous checkpoint is not released.
+- **`grammar` against `llm`.** The gain over the zero-shot base in FIM accuracy is +27.9 pp [+26.4, +29.5] on `grammar` and +31.8 pp [+29.9, +33.6] on `llm`; exec ÷ ceiling is 88.72% on `grammar` and 83.63% on `llm` (previous checkpoint: 82.80% and 66.09%). The gain of the v3 model over the previous checkpoint in FIM accuracy is +1.7 pp [+0.8, +2.7] on `grammar` and +17.8 pp [+16.3, +19.4] on `llm`. Both parts of the set resemble what the v3 corpus was built from (queries from the same LLM and grammar templates over other schemas), so these are not independent tests of style; the gain on LLM-written queries probably overstates the gain on queries written by people.
+- **Long cuts.** FIM accuracy on `stage` and `multiline` cuts is 8.06% and 11.04% (exact match against a long target within 96 new tokens); do not read the overall number as evenly spread over cut kinds.
+<!-- END:homolog-reading -->
+
+Limits of this evaluation: the authors are in the same family as the models (see above); the homologation is not fully independent of the v3 training data (see above); exec-valid means "runs", not "does what was intended"; the validator ceiling is 93.07% because 418 items refer to helpers (`ConnectionPool`, `getConnection`) that mongosh does not define; the smallest `cut_kind` has n=269 and the confidence intervals of the group tables are not shown; there is no multiplicity correction between groups; it is a single greedy run, with no variance across training runs.
+
+## Limitations and risks
+
+- **Synthetic data only.** Schemas are fictitious; the evaluation split uses unseen schemas of the same generator and templates. Expect lower scores on real collections, naming styles and query shapes.
+- **Invented field names.** It can propose fields, collections or operators that do not exist, and on unseen schemas the rate is not below the zero-shot base's in every setting (homologation section above, rates by `context_kind`). Validate field names before running a suggestion; do not rely on them unless the schema is in the context.
+- **The homologation numbers are not an estimate for real use**: the set is LLM- and grammar-written, not human-written, and is not fully independent of the v3 training data.
+- **A single training run per model** (one seed): differences between this model and the previous checkpoint are not backed by repeated runs.
+- **Dialects.** Trained on Mongo Console JavaScript (`ConnectionPool...` style) and JSON pipelines only. `mongosh` style prompts, other languages and chat are untested.
+- **No safety guarantee.** Behaviour around destructive operations (for example deletes or drops) was neither trained nor tested; during evaluation only read commands were executed and a guard blocked writes (0 write executions). The host application must keep a human in the loop and must not run suggestions blindly.
+- **Exec-valid is not correctness.** A validator-accepted completion may still not do what the user wanted; the validator itself accepts only 0.6255 of the references on the validation set.
+- **Autocomplete scope.** Short continuations (the IDE caps them at 32 tokens); long stage or multi-line targets are rarely matched exactly.
+- Not evaluated: bias or toxicity, long-context behaviour beyond 2048 tokens, multi-turn use, adversarial prompts, latency of the safetensors weights.
+
+## Formats
+
+| Format | Where | Notes |
+|---|---|---|
+| safetensors, bf16 | repository root | Reference weights, loaded with `transformers`. |
+
+This repository contains only the safetensors weights, the tokenizer and metadata (`provenance/`, `evaluation/`, `examples/`). ONNX builds for the editor are not part of it; they are planned as a separate repository (`esilva/KapiCoder-Mongo-Nano-ONNX`) and are not published yet. No GGUF is provided. Neither the teacher model nor the training data are part of this repository.
+
+## License, modification notice and attribution
+
+- License: **Apache License 2.0** (see `LICENSE`).
+- **Modification notice:** this model is a **modified** version of `Qwen/Qwen2.5-Coder-0.5B` (revision above). Changes: full fine-tuning (SFT) for MongoDB fill-in-the-middle autocomplete, with part of the labels distilled from a 7B teacher; export to safetensors bf16. It is not a product of Alibaba Cloud or the Qwen team and is not endorsed by them. See `NOTICE.md`.
+- Attribution: base model **Qwen2.5-Coder** by the Qwen team (Hui et al., *Qwen2.5-Coder Technical Report*, arXiv:2409.12186), Apache-2.0. The teacher was derived from Qwen2.5-Coder-7B (Apache-2.0); its weights are not part of this repository. Part of the training queries was written by Qwen3-14B (Apache-2.0).
+- The generic-code replay used npm packages under MIT, Apache-2.0, BSD-2/3-Clause and ISC licenses. The code itself is not redistributed; the package names, versions and licenses are listed in `REPLAY_ATTRIBUTION.md`.
+- The `editor-context-v1` prompt format is documented above so that others can build prompts for this model. Its use is not restricted by this card beyond the model license.
+
+Related model: [KapiCoder-Mongo-Small](https://huggingface.co/esilva/KapiCoder-Mongo-Small) (1.5B), same task and contract.
+
+---
+
+## Resumo em português (pt-BR)
+
+**Status: PREVIEW / experimental, preview público.** KapiCoder-Mongo-Nano é um modelo pequeno (cerca de 494 M de parâmetros) ajustado a partir do `Qwen/Qwen2.5-Coder-0.5B`
+para **completar consultas MongoDB no cursor** (preenchimento no meio, FIM) no editor Kapibara Studio. Esta é a versão **v3** dos pesos.
+
+- **O que é:** autocomplete. Sugere texto; **nunca executa consultas**. Não é chat nem agente, e não valida o que sugere.
+- **Uso:** prompt FIM do Qwen (`<|fim_prefix|>`, `<|fim_suffix|>`, `<|fim_middle|>`) com o cabeçalho de contexto `editor-context-v1` (veja o exemplo em inglês acima, que roda sem a IDE, também em `examples/example_fim.py`).
+- **Treino:** Ajuste completo (full fine-tuning), 2 épocas. 238.453 exemplos sintéticos (225.835 de MongoDB e 12.618 de replay de código JS/TS permissivo, 12,0% dos tokens), 64.999.846 tokens por época, 129.999.692 tokens vistos, 4.704 passos, 4,38 horas em uma RX 7800 XT. Os rótulos vêm em parte de um professor de 7B (mantidos só quando passam por validação de execução somente leitura), em parte de consultas escritas pelo Qwen3-14B e de uma gramática (rótulo de referência). Sem dados pessoais por construção; não houve revisão humana das amostras. **Uma única execução de treino (uma semente).**
+- **Avaliação:** **avaliação interna sobre dados sintéticos; sem conjunto cego humano.** As tabelas acima trazem os números reais (FIM acc, exec-valid, exec ÷ teto, campos inexistentes por `context_kind`, Generic-JS, base × modelo × checkpoint interno anterior não publicado). A homologação sintética fora da distribuição está no item seguinte; nada foi medido com consultas escritas por humanos.
+<!-- BEGIN:homolog-pt -->
+- **Homologação (avaliação interna, homologação sintética fora da distribuição):** 6.030 cortes de 28 schemas de telecom e adtech nunca vistos no treino, escritos pelo Qwen3-14B e por gramática, **não por humanos** (mesma família Qwen dos modelos). **A homologação não é totalmente independente do treino v3** (o corpus inclui consultas do mesmo gerador LLM e templates de gramática de outros schemas), então o ganho em consultas escritas pelo LLM provavelmente superestima o ganho em consultas humanas. FIM acc 42,34% (validação sintética: 40,50%; base: 12,62%; checkpoint interno anterior, não publicado: 33,10%); exec ÷ teto 86,17% (validação: 95,04%, com validadores diferentes). **Alerta:** com o schema visível (`full`), 10,81% das sugestões usam campos inexistentes, contra 7,02% da base zero-shot (validação sintética: 1,34%); sem schema, 9,09% (`irrelevant`) e 9,18% (`none`). Valide os nomes de campo antes de executar e não confie neles sem o schema no contexto. Executar não é acertar a intenção.
+<!-- END:homolog-pt -->
+- **Limites:** só dado sintético; em schemas novos há campos inexistentes mesmo com o schema visível (veja a homologação), e no conjunto de validação a taxa é maior sem schema visível; só dialetos Console e JSON; sem garantia de segurança. Valide os nomes de campo antes de executar.
+- **Formatos:** apenas safetensors bf16 neste repositório; o ONNX para a IDE será um repositório separado, ainda não publicado.
+- **Licença:** Apache-2.0, com **aviso de modificação**: modelo derivado e modificado do Qwen2.5-Coder-0.5B (equipe Qwen, Alibaba Cloud), sem endosso deles. O replay usa pacotes npm MIT/Apache-2.0/BSD/ISC, listados em `REPLAY_ATTRIBUTION.md`.

REPLAY_ATTRIBUTION.md

@@ -0,0 +1,582 @@
+# Replay attribution / Atribuição do replay
+
+A small share of the fine-tuning tokens (about 12%) is generic JavaScript/TypeScript taken from npm packages already installed on the
+training machine, used only as training text ("replay") to limit forgetting of generic code. The code is **not redistributed** with this model.
+All packages below are under permissive licenses (MIT, Apache-2.0, BSD-2/3-Clause, ISC, some dual-licensed with another permissive option); original headers were kept in the training text.
+Packages used only for evaluation are not listed. Source table: the project's `data/replay/v1/LICENSES.md`.
+
+Uma parcela pequena dos tokens de ajuste fino (cerca de 12%) é código JavaScript/TypeScript genérico de pacotes npm, usado só como texto de treino
+(replay) para limitar o esquecimento de código genérico. O código **não é redistribuído** com este modelo.
+
+Packages used in training: **555**.
+
+| License | Packages |
+|---|---:|
+| MIT | 402 |
+| ISC | 94 |
+| Apache-2.0 | 22 |
+| BSD-3-Clause | 21 |
+| BSD-2-Clause | 11 |
+| (Apache-2.0 AND BSD-3-Clause) | 1 |
+| (BSD-2-Clause OR MIT OR Apache-2.0) | 1 |
+| (MIT OR Apache-2.0) | 1 |
+| (MIT OR WTFPL) | 1 |
+| Apache 2.0 | 1 |
+
+| Package | Version | License | Copyright line |
+|---|---|---|---|
+| 7zip-bin | 5.2.0 | MIT | Copyright (c) 2016 Vladimir Krivosheev |
+| @ampproject/remapping | 2.2.1 | Apache-2.0 | - |
+| @azure/abort-controller | 2.1.2 | MIT | Copyright (c) 2020 Microsoft |
+| @azure/core-auth | 1.10.1 | MIT | Copyright (c) Microsoft Corporation. |
+| @azure/core-lro | 2.7.2 | MIT | Copyright (c) 2020 Microsoft |
+| @azure/core-paging | 1.6.2 | MIT | Copyright (c) 2020 Microsoft |
+| @azure/core-rest-pipeline | 1.24.0 | MIT | Copyright (c) Microsoft Corporation. |
+| @azure/core-tracing | 1.3.1 | MIT | Copyright (c) Microsoft Corporation. |
+| @azure/identity | 4.13.1 | MIT | Copyright (c) Microsoft Corporation. |
+| @azure/logger | 1.3.0 | MIT | Copyright (c) Microsoft Corporation. |
+| @azure/msal-browser | 5.13.0 | MIT | Copyright (c) Microsoft Corporation. All rights reserved. |
+| @azure/msal-common | 16.8.0 | MIT | Copyright (c) Microsoft Corporation. All rights reserved. |
+| @azure/msal-node | 5.2.4 | MIT | Copyright (c) 2020 Microsoft |
+| @azure/msal-node-runtime | 0.20.6 | MIT | - |
+| @azure/static-web-apps-cli | 2.0.9 | MIT | Copyright (c) 2021 Microsoft |
+| @babel/cli | 7.23.4 | MIT | Copyright (c) 2014-present Sebastian McKenzie and other contributors |
+| @babel/code-frame | 7.27.1 | MIT | Copyright (c) 2014-present Sebastian McKenzie and other contributors |
+| @babel/generator | 7.23.6 | MIT | Copyright (c) 2014-present Sebastian McKenzie and other contributors |
+| @babel/helper-annotate-as-pure | 7.22.5 | MIT | Copyright (c) 2014-present Sebastian McKenzie and other contributors |
+| @babel/helper-create-class-features-plugin | 7.23.5 | MIT | Copyright (c) 2014-present Sebastian McKenzie and other contributors |
+| @babel/helper-create-regexp-features-plugin | 7.22.15 | MIT | Copyright (c) 2014-present Sebastian McKenzie and other contributors |
+| @babel/helper-define-polyfill-provider | 0.4.4 | MIT | Copyright (c) 2014-present Nicolò Ribaudo and other contributors |
+| @babel/helper-environment-visitor | 7.22.20 | MIT | Copyright (c) 2014-present Sebastian McKenzie and other contributors |
+| @babel/helper-function-name | 7.23.0 | MIT | Copyright (c) 2014-present Sebastian McKenzie and other contributors |
+| @babel/helper-hoist-variables | 7.22.5 | MIT | Copyright (c) 2014-present Sebastian McKenzie and other contributors |
+| @babel/helper-member-expression-to-functions | 7.23.0 | MIT | Copyright (c) 2014-present Sebastian McKenzie and other contributors |
+| @babel/helper-optimise-call-expression | 7.22.5 | MIT | Copyright (c) 2014-present Sebastian McKenzie and other contributors |
+| @babel/helper-plugin-utils | 7.22.5 | MIT | Copyright (c) 2014-present Sebastian McKenzie and other contributors |
+| @babel/helper-remap-async-to-generator | 7.22.20 | MIT | Copyright (c) 2014-present Sebastian McKenzie and other contributors |
+| @babel/helper-replace-supers | 7.22.20 | MIT | Copyright (c) 2014-present Sebastian McKenzie and other contributors |
+| @babel/helper-simple-access | 7.22.5 | MIT | Copyright (c) 2014-present Sebastian McKenzie and other contributors |
+| @babel/helper-skip-transparent-expression-wrappers | 7.22.5 | MIT | Copyright (c) 2014-present Sebastian McKenzie and other contributors |
+| @babel/helper-split-export-declaration | 7.22.6 | MIT | Copyright (c) 2014-present Sebastian McKenzie and other contributors |
+| @babel/helper-string-parser | 7.27.1 | MIT | Copyright (c) 2014-present Sebastian McKenzie and other contributors |
+| @babel/helper-validator-identifier | 7.28.5 | MIT | Copyright (c) 2014-present Sebastian McKenzie and other contributors |
+| @babel/helper-validator-option | 7.23.5 | MIT | Copyright (c) 2014-present Sebastian McKenzie and other contributors |
+| @babel/helper-wrap-function | 7.22.20 | MIT | Copyright (c) 2014-present Sebastian McKenzie and other contributors |
+| @babel/helpers | 7.28.4 | MIT | Copyright (c) 2014-present Sebastian McKenzie and other contributors |
+| @babel/parser | 7.28.5 | MIT | Copyright (C) 2012-2014 by various contributors (see AUTHORS) |
+| @babel/plugin-bugfix-safari-id-destructuring-collision-in-function-expression | 7.23.3 | MIT | Copyright (c) 2014-present Sebastian McKenzie and other contributors |
+| @babel/plugin-bugfix-v8-spread-parameters-in-optional-chaining | 7.23.3 | MIT | Copyright (c) 2014-present Sebastian McKenzie and other contributors |
+| @babel/plugin-bugfix-v8-static-class-fields-redefine-readonly | 7.23.7 | MIT | Copyright (c) 2014-present Sebastian McKenzie and other contributors |
+| @babel/plugin-proposal-private-property-in-object | 7.21.0-placeholder-for-preset-env.2 | MIT | Copyright (c) 2014-present Sebastian McKenzie and other contributors |
+| @babel/plugin-syntax-async-generators | 7.8.4 | MIT | Copyright (c) 2014-present Sebastian McKenzie and other contributors |
+| @babel/plugin-syntax-class-properties | 7.12.13 | MIT | Copyright (c) 2014-present Sebastian McKenzie and other contributors |
+| @babel/plugin-syntax-class-static-block | 7.14.5 | MIT | Copyright (c) 2014-present Sebastian McKenzie and other contributors |
+| @babel/plugin-syntax-dynamic-import | 7.8.3 | MIT | Copyright (c) 2014-present Sebastian McKenzie and other contributors |
+| @babel/plugin-syntax-export-namespace-from | 7.8.3 | MIT | Copyright (c) 2014-present Sebastian McKenzie and other contributors |
+| @babel/plugin-syntax-import-assertions | 7.23.3 | MIT | Copyright (c) 2014-present Sebastian McKenzie and other contributors |
+| @babel/plugin-syntax-import-attributes | 7.23.3 | MIT | Copyright (c) 2014-present Sebastian McKenzie and other contributors |
+| @babel/plugin-syntax-import-meta | 7.10.4 | MIT | Copyright (c) 2014-present Sebastian McKenzie and other contributors |
+| @babel/plugin-syntax-json-strings | 7.8.3 | MIT | Copyright (c) 2014-present Sebastian McKenzie and other contributors |
+| @babel/plugin-syntax-logical-assignment-operators | 7.10.4 | MIT | Copyright (c) 2014-present Sebastian McKenzie and other contributors |
+| @babel/plugin-syntax-nullish-coalescing-operator | 7.8.3 | MIT | Copyright (c) 2014-present Sebastian McKenzie and other contributors |
+| @babel/plugin-syntax-numeric-separator | 7.10.4 | MIT | Copyright (c) 2014-present Sebastian McKenzie and other contributors |
+| @babel/plugin-syntax-object-rest-spread | 7.8.3 | MIT | Copyright (c) 2014-present Sebastian McKenzie and other contributors |
+| @babel/plugin-syntax-optional-catch-binding | 7.8.3 | MIT | Copyright (c) 2014-present Sebastian McKenzie and other contributors |
+| @babel/plugin-syntax-optional-chaining | 7.8.3 | MIT | Copyright (c) 2014-present Sebastian McKenzie and other contributors |
+| @babel/plugin-syntax-private-property-in-object | 7.14.5 | MIT | Copyright (c) 2014-present Sebastian McKenzie and other contributors |
+| @babel/plugin-syntax-top-level-await | 7.14.5 | MIT | Copyright (c) 2014-present Sebastian McKenzie and other contributors |
+| @babel/plugin-syntax-unicode-sets-regex | 7.18.6 | MIT | Copyright (c) 2014-present Sebastian McKenzie and other contributors |
+| @babel/plugin-transform-arrow-functions | 7.23.3 | MIT | Copyright (c) 2014-present Sebastian McKenzie and other contributors |
+| @babel/plugin-transform-async-generator-functions | 7.23.7 | MIT | Copyright (c) 2014-present Sebastian McKenzie and other contributors |
+| @babel/plugin-transform-async-to-generator | 7.23.3 | MIT | Copyright (c) 2014-present Sebastian McKenzie and other contributors |
+| @babel/plugin-transform-block-scoped-functions | 7.23.3 | MIT | Copyright (c) 2014-present Sebastian McKenzie and other contributors |
+| @babel/plugin-transform-class-properties | 7.23.3 | MIT | Copyright (c) 2014-present Sebastian McKenzie and other contributors |
+| @babel/plugin-transform-class-static-block | 7.23.4 | MIT | Copyright (c) 2014-present Sebastian McKenzie and other contributors |
+| @babel/plugin-transform-classes | 7.23.8 | MIT | Copyright (c) 2014-present Sebastian McKenzie and other contributors |
+| @babel/plugin-transform-computed-properties | 7.23.3 | MIT | Copyright (c) 2014-present Sebastian McKenzie and other contributors |
+| @babel/plugin-transform-destructuring | 7.23.3 | MIT | Copyright (c) 2014-present Sebastian McKenzie and other contributors |
+| @babel/plugin-transform-dotall-regex | 7.23.3 | MIT | Copyright (c) 2014-present Sebastian McKenzie and other contributors |
+| @babel/plugin-transform-duplicate-keys | 7.23.3 | MIT | Copyright (c) 2014-present Sebastian McKenzie and other contributors |
+| @babel/plugin-transform-dynamic-import | 7.23.4 | MIT | Copyright (c) 2014-present Sebastian McKenzie and other contributors |
+| @babel/plugin-transform-exponentiation-operator | 7.23.3 | MIT | Copyright (c) 2014-present Sebastian McKenzie and other contributors |
+| @babel/plugin-transform-export-namespace-from | 7.23.4 | MIT | Copyright (c) 2014-present Sebastian McKenzie and other contributors |
+| @babel/plugin-transform-for-of | 7.23.6 | MIT | Copyright (c) 2014-present Sebastian McKenzie and other contributors |
+| @babel/plugin-transform-function-name | 7.23.3 | MIT | Copyright (c) 2014-present Sebastian McKenzie and other contributors |
+| @babel/plugin-transform-json-strings | 7.23.4 | MIT | Copyright (c) 2014-present Sebastian McKenzie and other contributors |
+| @babel/plugin-transform-literals | 7.23.3 | MIT | Copyright (c) 2014-present Sebastian McKenzie and other contributors |
+| @babel/plugin-transform-logical-assignment-operators | 7.23.4 | MIT | Copyright (c) 2014-present Sebastian McKenzie and other contributors |
+| @babel/plugin-transform-member-expression-literals | 7.23.3 | MIT | Copyright (c) 2014-present Sebastian McKenzie and other contributors |
+| @babel/plugin-transform-modules-amd | 7.23.3 | MIT | Copyright (c) 2014-present Sebastian McKenzie and other contributors |
+| @babel/plugin-transform-modules-commonjs | 7.23.3 | MIT | Copyright (c) 2014-present Sebastian McKenzie and other contributors |
+| @babel/plugin-transform-modules-systemjs | 7.23.3 | MIT | Copyright (c) 2014-present Sebastian McKenzie and other contributors |
+| @babel/plugin-transform-modules-umd | 7.23.3 | MIT | Copyright (c) 2014-present Sebastian McKenzie and other contributors |
+| @babel/plugin-transform-named-capturing-groups-regex | 7.22.5 | MIT | Copyright (c) 2014-present Sebastian McKenzie and other contributors |
+| @babel/plugin-transform-new-target | 7.23.3 | MIT | Copyright (c) 2014-present Sebastian McKenzie and other contributors |
+| @babel/plugin-transform-nullish-coalescing-operator | 7.23.4 | MIT | Copyright (c) 2014-present Sebastian McKenzie and other contributors |
+| @babel/plugin-transform-numeric-separator | 7.23.4 | MIT | Copyright (c) 2014-present Sebastian McKenzie and other contributors |
+| @babel/plugin-transform-object-rest-spread | 7.23.4 | MIT | Copyright (c) 2014-present Sebastian McKenzie and other contributors |
+| @babel/plugin-transform-object-super | 7.23.3 | MIT | Copyright (c) 2014-present Sebastian McKenzie and other contributors |
+| @babel/plugin-transform-optional-catch-binding | 7.23.4 | MIT | Copyright (c) 2014-present Sebastian McKenzie and other contributors |
+| @babel/plugin-transform-optional-chaining | 7.23.4 | MIT | Copyright (c) 2014-present Sebastian McKenzie and other contributors |
+| @babel/plugin-transform-parameters | 7.23.3 | MIT | Copyright (c) 2014-present Sebastian McKenzie and other contributors |
+| @babel/plugin-transform-private-methods | 7.23.3 | MIT | Copyright (c) 2014-present Sebastian McKenzie and other contributors |
+| @babel/plugin-transform-private-property-in-object | 7.23.4 | MIT | Copyright (c) 2014-present Sebastian McKenzie and other contributors |
+| @babel/plugin-transform-property-literals | 7.23.3 | MIT | Copyright (c) 2014-present Sebastian McKenzie and other contributors |
+| @babel/plugin-transform-regenerator | 7.23.3 | MIT | Copyright (c) 2014-present Sebastian McKenzie and other contributors |
+| @babel/plugin-transform-reserved-words | 7.23.3 | MIT | Copyright (c) 2014-present Sebastian McKenzie and other contributors |
+| @babel/plugin-transform-shorthand-properties | 7.23.3 | MIT | Copyright (c) 2014-present Sebastian McKenzie and other contributors |
+| @babel/plugin-transform-spread | 7.23.3 | MIT | Copyright (c) 2014-present Sebastian McKenzie and other contributors |
+| @babel/plugin-transform-sticky-regex | 7.23.3 | MIT | Copyright (c) 2014-present Sebastian McKenzie and other contributors |
+| @babel/plugin-transform-template-literals | 7.23.3 | MIT | Copyright (c) 2014-present Sebastian McKenzie and other contributors |
+| @babel/plugin-transform-typeof-symbol | 7.23.3 | MIT | Copyright (c) 2014-present Sebastian McKenzie and other contributors |
+| @babel/plugin-transform-unicode-escapes | 7.23.3 | MIT | Copyright (c) 2014-present Sebastian McKenzie and other contributors |
+| @babel/plugin-transform-unicode-property-regex | 7.23.3 | MIT | Copyright (c) 2014-present Sebastian McKenzie and other contributors |
+| @babel/plugin-transform-unicode-regex | 7.23.3 | MIT | Copyright (c) 2014-present Sebastian McKenzie and other contributors |
+| @babel/plugin-transform-unicode-sets-regex | 7.23.3 | MIT | Copyright (c) 2014-present Sebastian McKenzie and other contributors |
+| @babel/preset-env | 7.23.8 | MIT | Copyright (c) 2014-present Sebastian McKenzie and other contributors |
+| @babel/preset-modules | 0.1.6-no-external-plugins | MIT | Copyright (c) 2020 Babel |
+| @babel/regjsgen | 0.8.0 | MIT | Copyright 2014-2020 Benjamin Tan <https://ofcr.se/> |
+| @babel/runtime | 7.29.7 | MIT | Copyright (c) 2014-present Sebastian McKenzie and other contributors |
+| @babel/traverse | 7.23.7 | MIT | Copyright (c) 2014-present Sebastian McKenzie and other contributors |
+| @bufbuild/protobuf | 2.13.0 | (Apache-2.0 AND BSD-3-Clause) | - |
+| @gar/promise-retry | 1.0.3 | MIT | Copyright (c) 2011 Tim Koschützki, Felix Geisendörfer |
+| @grpc/grpc-js | 1.9.0 | Apache-2.0 | - |
+| @hapi/topo | 5.1.0 | BSD-3-Clause | Copyright (c) 2012-2020, Sideway Inc, and project contributors |
+| @img/colour | 1.1.0 | MIT | Copyright (c) 2012 Heather Arthur |
+| @isaacs/cliui | 8.0.2 | ISC | Copyright (c) 2015, Contributors |
+| @isaacs/string-locale-compare | 1.1.0 | ISC | Copyright (c) Isaac Z. Schlueter |
+| @jridgewell/gen-mapping | 0.3.3 | MIT | Copyright 2022 Justin Ridgewell |
+| @jridgewell/resolve-uri | 3.1.1 | MIT | Copyright 2019 Justin Ridgewell |
+| @jridgewell/set-array | 1.1.2 | MIT | Copyright 2022 Justin Ridgewell |
+| @jridgewell/source-map | 0.3.5 | MIT | Copyright 2019 Justin Ridgewell |
+| @jridgewell/sourcemap-codec | 1.4.15 | MIT | Copyright (c) 2015 Rich Harris |
+| @jridgewell/trace-mapping | 0.3.31 | MIT | Copyright 2024 Justin Ridgewell |
+| @lmstudio/lms-isomorphic | 0.4.6 | Apache-2.0 | Copyright (c) 2025 Element Labs Inc |
+| @npmcli/agent | 4.0.0 | ISC | - |
+| @npmcli/arborist | 9.4.3 | ISC | Copyright npm, Inc. |
+| @npmcli/config | 10.8.1 | ISC | Copyright (c) npm, Inc. |
+| @npmcli/fs | 5.0.0 | ISC | Copyright npm, Inc. |
+| @npmcli/git | 7.0.2 | ISC | Copyright (c) npm, Inc. |
+| @npmcli/installed-package-contents | 4.0.0 | ISC | Copyright (c) npm, Inc. |
+| @npmcli/map-workspaces | 5.0.3 | ISC | Copyright npm, Inc. |
+| @npmcli/node-gyp | 5.0.0 | ISC | Copyright (c) 2023 by GitHub Inc. |
+| @npmcli/package-json | 7.0.5 | ISC | Copyright GitHub Inc. |
+| @npmcli/query | 5.0.0 | ISC | Copyright npm, Inc. |
+| @npmcli/redact | 4.0.0 | ISC | Copyright (c) 2024 npm |
+| @npmcli/run-script | 10.0.4 | ISC | Copyright (c) npm, Inc. |
+| @picovoice/pvrecorder-node | 1.2.9 | Apache-2.0 | - |
+| @pnpm/config.env-replace | 1.1.0 | MIT | - |
+| @pnpm/network.ca-file | 1.0.2 | MIT | - |
+| @pnpm/npm-conf | 3.0.3 | MIT | Copyright (c) Kevin Mårtensson (github.com/kevva) |
+| @pondwader/socks5-server | 1.0.10 | MIT | Copyright (c) 2024 PondWader |
+| @protobufjs/aspromise | 1.1.2 | BSD-3-Clause | Copyright (c) 2016, Daniel Wirtz  All rights reserved. |
+| @protobufjs/base64 | 1.1.2 | BSD-3-Clause | Copyright (c) 2016, Daniel Wirtz  All rights reserved. |
+| @protobufjs/codegen | 2.0.5 | BSD-3-Clause | Copyright (c) 2016, Daniel Wirtz  All rights reserved. |
+| @protobufjs/eventemitter | 1.1.1 | BSD-3-Clause | Copyright (c) 2016, Daniel Wirtz  All rights reserved. |
+| @protobufjs/fetch | 1.1.1 | BSD-3-Clause | Copyright (c) 2016, Daniel Wirtz  All rights reserved. |
+| @protobufjs/path | 1.1.2 | BSD-3-Clause | Copyright (c) 2016, Daniel Wirtz  All rights reserved. |
+| @protobufjs/pool | 1.1.0 | BSD-3-Clause | Copyright (c) 2016, Daniel Wirtz  All rights reserved. |
+| @protobufjs/utf8 | 1.1.2 | BSD-3-Clause | Copyright (c) 2016, Daniel Wirtz  All rights reserved. |
+| @sagold/json-pointer | 5.1.2 | MIT | Copyright 2022 Sascha Goldhofer |
+| @sagold/json-query | 6.2.0 | MIT | Copyright 2022 Sascha Goldhofer |
+| @sideway/address | 4.1.5 | BSD-3-Clause | Copyright (c) 2019-2020, Sideway, Inc. and Project contributors |
+| @sideway/pinpoint | 2.0.0 | BSD-3-Clause | Copyright (c) 2019-2020, Sideway. Inc, and project contributors |
+| @sigstore/core | 3.2.0 | Apache-2.0 | - |
+| @sigstore/protobuf-specs | 0.5.1 | Apache-2.0 | - |
+| @sigstore/sign | 4.1.1 | Apache-2.0 | - |
+| @sigstore/verify | 3.1.0 | Apache-2.0 | - |
+| @statsig/client-core | 3.33.3 | ISC | - |
+| @statsig/js-client | 3.33.3 | ISC | Copyright (c) 2021, Statsig, Inc. |
+| @tufjs/canonical-json | 2.0.0 | MIT | Copyright (c) 2022 GitHub and the TUF Contributors |
+| @typespec/ts-http-runtime | 0.3.6 | MIT | Copyright (c) Microsoft Corporation. |
+| @vscode/fs-copyfile | 2.0.0 | MIT | - |
+| @vscode/ripgrep | 1.15.9 | MIT | Copyright (c) Microsoft Corporation |
+| @vscode/sandbox-runtime | 0.0.1 | Apache-2.0 | - |
+| @webviewjs/webview | 0.4.0 | MIT | Copyright (c) 2024 Twilight |
+| abbrev | 4.0.0 | ISC | Copyright (c) Isaac Z. Schlueter and Contributors |
+| abstract-level | 3.1.1 | MIT | Copyright © 2013 Rod Vagg and the contributors to abstract-level. |
+| acorn | 8.15.0 | MIT | Copyright (C) 2012-2022 by various contributors (see AUTHORS) |
+| agent-base | 7.1.4 | MIT | Copyright (c) 2013 Nathan Rajlich |
+| ansi-align | 3.0.1 | ISC | Copyright (c) 2016, Contributors |
+| ansi-escapes | 4.3.2 | MIT | Copyright (c) Sindre Sorhus (https://sindresorhus.com) |
+| ansi-regex | 6.2.2 | MIT | Copyright (c) Sindre Sorhus (https://sindresorhus.com) |
+| ansi-styles | 6.2.3 | MIT | Copyright (c) Sindre Sorhus (https://sindresorhus.com) |
+| anymatch | 3.1.3 | ISC | Copyright (c) 2019 Elan Shanker, Paul Miller (https://paulmillr.com) |
+| application-config-path | 0.1.1 | MIT | Copyright (c) 2015 Linus Unnebäck |
+| aproba | 2.1.0 | ISC | Copyright (c) 2015, Rebecca Turner |
+| asynckit | 0.4.0 | MIT | Copyright (c) 2016 Alex Indigo |
+| atomically | 2.1.1 | MIT | Copyright (c) 2020-present Fabio Spampinato |
+| axios | 1.17.0 | MIT | - |
+| azure-functions-nodejs-worker | 3.14.1 | (MIT OR Apache-2.0) | Copyright (c) .NET Foundation. All rights reserved. |
+| babel-plugin-polyfill-regenerator | 0.5.4 | MIT | Copyright (c) 2014-present Nicolò Ribaudo and other contributors |
+| balanced-match | 1.0.2 | MIT | Copyright (c) 2013 Julian Gruber &lt;&gt; |
+| base64-js | 1.5.1 | MIT | Copyright (c) 2014 Jameson Little |
+| bin-links | 6.0.0 | ISC | Copyright (c) npm, Inc. |
+| bl | 4.1.0 | MIT | Copyright (c) 2013-2019 bl contributors |
+| boxen | 8.0.1 | MIT | Copyright (c) Sindre Sorhus (https://sindresorhus.com) |
+| brace-expansion | 5.0.5 | MIT | Copyright Julian Gruber |
+| braces | 3.0.3 | MIT | Copyright (c) 2014-present, Jon Schlinkert. |
+| buffer | 5.7.1 | MIT | Copyright (c) Feross Aboukhadijeh, and other contributors. |
+| buffer-equal-constant-time | 1.0.1 | BSD-3-Clause | Copyright (c) 2013, GoInstant Inc., a salesforce.com company |
+| buffer-from | 1.1.2 | MIT | Copyright (c) 2016, 2018 Linus Unnebäck |
+| bufferutil | 4.0.1 | MIT | Copyright (c) 2011 Einar Otto Stangvik (http://2x.io) |
+| call-bind-apply-helpers | 1.0.2 | MIT | Copyright (c) 2024 Jordan Harband |
+| camelcase | 8.0.0 | MIT | Copyright (c) Sindre Sorhus (https://sindresorhus.com) |
+| chalk | 5.6.2 | MIT | Copyright (c) Sindre Sorhus (https://sindresorhus.com) |
+| charenc | 0.0.2 | BSD-3-Clause | - |
+| chokidar | 3.5.3 | MIT | Copyright (c) 2012-2019 Paul Miller (https://paulmillr.com), Elan Shanker |
+| chownr | 1.1.4 | ISC | Copyright (c) Isaac Z. Schlueter and Contributors |
+| chrome-devtools-mcp | 1.1.0 | Apache-2.0 | - |
+| ci-info | 4.4.0 | MIT | Copyright (c) 2016 Thomas Watson Steen |
+| classic-level | 3.0.0 | MIT | Copyright (c) 2012 Rod Vagg and the contributors to classic-level and leveldown. |
+| clean-css | 5.3.3 | MIT | Copyright (C) 2017 JakubPawlowicz.com |
+| cli-cursor | 3.1.0 | MIT | Copyright (c) Sindre Sorhus (sindresorhus.com) |
+| cli-spinners | 2.9.2 | MIT | Copyright (c) Sindre Sorhus (https://sindresorhus.com) |
+| cliui | 8.0.1 | ISC | Copyright (c) 2015, Contributors |
+| clone | 2.1.2 | MIT | Copyright © 2011-2015 Paul Vorbach |
+| clone-buffer | 1.0.0 | MIT | Copyright (c) 2016 Blaine Bublitz, Eric Schoffstall and other contributors |
+| clone-stats | 1.0.0 | MIT | Copyright (c) 2014 Hugh Kennedy |
+| cloneable-readable | 1.1.3 | MIT | Copyright (c) 2016 Matteo Collina |
+| color-convert | 2.0.1 | MIT | Copyright (c) 2011-2016 Heather Arthur |
+| color-name | 1.1.4 | MIT | Copyright (c) 2015 Dmitry Ivanov |
+| combined-stream | 1.0.8 | MIT | Copyright (c) 2011 Debuggable Limited |
+| command-exists | 1.2.9 | MIT | Copyright (c) 2014 Matthew Conlen |
+| concurrently | 7.6.0 | MIT | Copyright (c) 2015 Kimmo Brunfeldt |
+| config-chain | 1.1.13 | MIT | Copyright (c) 2011 Dominic Tarr |
+| configstore | 7.1.0 | BSD-2-Clause | Copyright (c) Google |
+| continue | 2.0.0 | Apache-2.0 | Copyright 2023 Continue |
+| cookie | 0.7.2 | MIT | Copyright (c) 2012-2014 Roman Shtylman |
+| core-util-is | 1.0.3 | MIT | Copyright Node.js contributors. All rights reserved. |
+| cross-spawn | 7.0.6 | MIT | Copyright (c) 2018 Made With MOXY Lda |
+| crypt | 0.0.2 | BSD-3-Clause | - |
+| cssesc | 3.0.0 | MIT | Copyright Mathias Bynens <https://mathiasbynens.be/> |
+| decompress-response | 6.0.0 | MIT | Copyright (c) Sindre Sorhus (https://sindresorhus.com) |
+| deep-extend | 0.6.0 | MIT | Copyright (c) 2013-2018, Viacheslav Lotsmanov |
+| deepmerge | 4.3.1 | MIT | Copyright (c) 2012 James Halliday, Josh Duff, and other contributors |
+| default-browser | 5.5.0 | MIT | Copyright (c) Sindre Sorhus (https://sindresorhus.com) |
+| default-browser-id | 5.0.1 | MIT | Copyright (c) Sindre Sorhus (https://sindresorhus.com) |
+| default-gateway | 6.0.3 | BSD-2-Clause | Copyright (c) silverwind |
+| defaults | 1.0.4 | MIT | Copyright (c) 2022 Sindre Sorhus |
+| define-lazy-prop | 3.0.0 | MIT | Copyright (c) Sindre Sorhus (https://sindresorhus.com) |
+| delayed-stream | 1.0.0 | MIT | Copyright (c) 2011 Debuggable Limited |
+| depd | 2.0.0 | MIT | Copyright (c) 2014-2018 Douglas Christopher Wilson |
+| destroy | 1.2.0 | MIT | Copyright (c) 2014 Jonathan Ong |
+| detect-libc | 2.1.2 | Apache-2.0 | - |
+| devcert | 1.2.3 | MIT | - |
+| discontinuous-range | 1.0.0 | MIT | Copyright (c) 2014 David Tudury |
+| dot-case | 3.0.4 | MIT | Copyright (c) 2014 Blake Embrey |
+| dot-prop | 9.0.0 | MIT | Copyright (c) Sindre Sorhus (https://sindresorhus.com) |
+| dotenv | 16.6.1 | BSD-2-Clause | Copyright (c) 2015, Scott Motte |
+| dunder-proto | 1.0.1 | MIT | Copyright (c) 2024 ECMAScript Shims |
+| eastasianwidth | 0.2.0 | MIT | - |
+| ebnf | 1.9.1 | MIT | Copyright (c) 2016 Agustin Mendez |
+| ee-first | 1.1.1 | MIT | Copyright (c) 2014 Jonathan Ong |
+| electron-to-chromium | 1.5.250 | ISC | Copyright 2018 Kilian Valkhof |
+| encodeurl | 2.0.0 | MIT | Copyright (c) 2016 Douglas Christopher Wilson |
+| entities | 4.5.0 | BSD-2-Clause | Copyright (c) Felix Böhm |
+| env-paths | 2.2.1 | MIT | Copyright (c) Sindre Sorhus (sindresorhus.com) |
+| eol | 0.9.1 | MIT | - |
+| es-define-property | 1.0.1 | MIT | Copyright (c) 2024 Jordan Harband |
+| es-errors | 1.3.0 | MIT | Copyright (c) 2024 Jordan Harband |
+| es-object-atoms | 1.1.2 | MIT | Copyright (c) 2024 Jordan Harband |
+| es-set-tostringtag | 2.1.0 | MIT | Copyright (c) 2022 ECMAScript Shims |
+| es6-iterator | 2.0.3 | MIT | Copyright (C) 2013-2017 Mariusz Nowak (www.medikoo.com) |
+| es6-symbol | 3.1.3 | ISC | Copyright (c) 2013-2019, Mariusz Nowak, @medikoo, medikoo.com |
+| escalade | 3.2.0 | MIT | Copyright (c) Luke Edwards (lukeed.com) |
+| escape-goat | 4.0.0 | MIT | Copyright (c) Sindre Sorhus (https://sindresorhus.com) |
+| escape-html | 1.0.3 | MIT | Copyright (c) 2012-2013 TJ Holowaychuk |
+| esutils | 2.0.3 | BSD-2-Clause | - |
+| etag | 1.8.1 | MIT | Copyright (c) 2014-2016 Douglas Christopher Wilson |
+| eventemitter3 | 4.0.7 | MIT | Copyright (c) 2014 Arnout Kazemier |
+| execa | 5.1.1 | MIT | Copyright (c) Sindre Sorhus (https://sindresorhus.com) |
+| expand-template | 2.0.3 | (MIT OR WTFPL) | Copyright (c) 2018 Lars-Magnus Skog |
+| ext | 1.4.0 | ISC | Copyright (c) 2011-2019, Mariusz Nowak, @medikoo, medikoo.com |
+| fast-copy | 3.0.2 | MIT | Copyright (c) 2018 Tony Quetano |
+| fastest-levenshtein | 1.0.16 | MIT | Copyright (c) 2020 Kasper Unn Weihe |
+| fdir | 6.5.0 | MIT | Copyright 2023 Abdullah Atta |
+| file-uri-to-path | 1.0.0 | MIT | Copyright (c) 2014 Nathan Rajlich |
+| fill-range | 7.1.1 | MIT | Copyright (c) 2014-present, Jon Schlinkert. |
+| finalhandler | 1.3.2 | MIT | Copyright (c) 2014-2022 Douglas Christopher Wilson |
+| follow-redirects | 1.16.0 | MIT | Copyright 2014–present Olivier Lalonde, James Talmage, Ruben Verborgh |
+| foreground-child | 3.3.1 | ISC | Copyright (c) 2015-2023 Isaac Z. Schlueter and Contributors |
+| form-data | 4.0.5 | MIT | Copyright (c) 2012 Felix Geisendörfer and contributors |
+| foundry-local-sdk | 1.2.3 | MIT | - |
+| fresh | 0.5.2 | MIT | Copyright (c) 2012 TJ Holowaychuk |
+| fs-minipass | 3.0.3 | ISC | Copyright (c) Isaac Z. Schlueter and Contributors |
+| fs-readdir-recursive | 1.1.0 | MIT | Copyright (c) 2014 Jonathan Ong |
+| fs.realpath | 1.0.0 | ISC | Copyright (c) Isaac Z. Schlueter and Contributors |
+| function-bind | 1.1.2 | MIT | Copyright (c) 2013 Raynos. |
+| gensync | 1.0.0-beta.2 | MIT | Copyright 2018 Logan Smyth |
+| get-caller-file | 2.0.5 | ISC | Copyright 2018 Stefan Penner |
+| get-east-asian-width | 1.6.0 | MIT | Copyright (c) Sindre Sorhus (https://sindresorhus.com) |
+| get-intrinsic | 1.3.0 | MIT | Copyright (c) 2020 Jordan Harband |
+| get-port | 5.1.1 | MIT | Copyright (c) Sindre Sorhus (sindresorhus.com) |
+| get-proto | 1.0.1 | MIT | Copyright (c) 2025 Jordan Harband |
+| get-stream | 6.0.1 | MIT | Copyright (c) Sindre Sorhus (https://sindresorhus.com) |
+| github-from-package | 0.0.0 | MIT | - |
+| glob | 10.5.0 | ISC | Copyright (c) 2009-2023 Isaac Z. Schlueter and Contributors |
+| glob-parent | 5.1.2 | ISC | Copyright (c) 2015, 2019 Elan Shanker |
+| global-directory | 4.0.1 | MIT | Copyright (c) Sindre Sorhus (https://sindresorhus.com) |
+| gopd | 1.2.0 | MIT | Copyright (c) 2022 Jordan Harband |
+| graceful-fs | 4.2.11 | ISC | Copyright (c) 2011-2022 Isaac Z. Schlueter, Ben Noordhuis, and Contributors |
+| handlebars | 4.7.9 | MIT | Copyright (C) 2011-2019 by Yehuda Katz |
+| has-symbols | 1.1.0 | MIT | Copyright (c) 2016 Jordan Harband |
+| has-tostringtag | 1.0.2 | MIT | Copyright (c) 2021 Inspect JS |
+| hosted-git-info | 9.0.2 | ISC | Copyright (c) 2015, Rebecca Turner |
+| html-minifier-terser | 7.2.0 | MIT | Copyright (c) 2010-2019 Juriy "kangax" Zaytsev |
+| http-cache-semantics | 4.2.0 | BSD-2-Clause | Copyright 2016-2018 Kornel Lesiński |
+| http-errors | 2.0.1 | MIT | Copyright (c) 2014 Jonathan Ong |
+| http-proxy-agent | 7.0.2 | MIT | Copyright (c) 2013 Nathan Rajlich |
+| human-signals | 2.1.0 | Apache-2.0 | - |
+| iconv-lite | 0.7.2 | MIT | Copyright (c) 2011 Alexander Shtuchkin |
+| ieee754 | 1.2.1 | BSD-3-Clause | Copyright 2008 Fair Oaks Labs, Inc. |
+| ignore-walk | 8.0.0 | ISC | Copyright (c) Isaac Z. Schlueter and Contributors |
+| inflight | 1.0.6 | ISC | Copyright (c) Isaac Z. Schlueter |
+| inherits | 2.0.4 | ISC | Copyright (c) Isaac Z. Schlueter |
+| ini | 6.0.0 | ISC | Copyright (c) Isaac Z. Schlueter and Contributors |
+| init-package-json | 8.2.5 | ISC | Copyright npm, Inc. |
+| internal-ip | 6.2.0 | MIT | Copyright (c) Sindre Sorhus (https://sindresorhus.com) |
+| ip-address | 10.1.0 | MIT | Copyright (C) 2011 by Beau Gunderson |
+| ipaddr.js | 1.9.1 | MIT | Copyright (C) 2011-2017 whitequark |
+| is-buffer | 1.1.6 | MIT | Copyright (c) Feross Aboukhadijeh |
+| is-core-module | 2.13.1 | MIT | Copyright (c) 2014 Dave Justice |
+| is-docker | 3.0.0 | MIT | Copyright (c) Sindre Sorhus (https://sindresorhus.com) |
+| is-extglob | 2.1.1 | MIT | Copyright (c) 2014-2016, Jon Schlinkert |
+| is-fullwidth-code-point | 3.0.0 | MIT | Copyright (c) Sindre Sorhus (sindresorhus.com) |
+| is-glob | 4.0.3 | MIT | Copyright (c) 2014-2017, Jon Schlinkert. |
+| is-inside-container | 1.0.0 | MIT | Copyright (c) Sindre Sorhus (https://sindresorhus.com) |
+| is-installed-globally | 1.0.0 | MIT | Copyright (c) Sindre Sorhus (https://sindresorhus.com) |
+| is-number | 7.0.0 | MIT | Copyright (c) 2014-present, Jon Schlinkert. |
+| is-stream | 2.0.1 | MIT | Copyright (c) Sindre Sorhus (https://sindresorhus.com) |
+| is-typedarray | 1.0.0 | MIT | - |
+| is-valid-domain | 0.1.6 | MIT | Copyright (C) 2014 Miguel Mota |
+| is-wsl | 3.1.1 | MIT | Copyright (c) Sindre Sorhus (https://sindresorhus.com) |
+| isarray | 1.0.0 | MIT | - |
+| isexe | 2.0.0 | ISC | Copyright (c) Isaac Z. Schlueter and Contributors |
+| joi | 17.13.4 | BSD-3-Clause | Copyright (c) 2012-2022, Project contributors. |
+| jose | 6.2.3 | MIT | Copyright (c) 2018 Filip Skokan |
+| jpeg-js | 0.4.4 | BSD-3-Clause | Copyright (c) 2014, Eugene Ware |
+| jsesc | 2.5.2 | MIT | Copyright Mathias Bynens <https://mathiasbynens.be/> |
+| json-parse-even-better-errors | 5.0.0 | MIT | Copyright 2017 Kat Marchán |
+| json-schema-library | 9.3.5 | MIT | Copyright 2022 Sascha Goldhofer |
+| json-stringify-nice | 1.1.4 | ISC | Copyright (c) Isaac Z. Schlueter and Contributors |
+| json5 | 2.2.3 | MIT | Copyright (c) 2012-2018 Aseem Kishore, and [others]. |
+| jsonparse | 1.3.1 | MIT | Copyright (c) 2012 Tim Caswell |
+| jsonschema | 1.5.0 | MIT | Copyright (C) 2012-2015 Tom de Grunt |
+| just-diff | 6.0.2 | MIT | Copyright (c) 2016 angus croll |
+| just-diff-apply | 5.5.0 | MIT | Copyright (c) 2016 angus croll |
+| jwa | 2.0.1 | MIT | Copyright (c) 2013 Brian J. Brennan |
+| jws | 4.0.1 | MIT | Copyright (c) 2013 Brian J. Brennan |
+| jwt-decode | 4.0.0 | MIT | Copyright (c) 2015 Auth0, Inc. (http://auth0.com) |
+| keytar | 7.9.0 | MIT | Copyright (c) 2013 GitHub Inc. |
+| kleur | 3.0.3 | MIT | Copyright (c) Luke Edwards (lukeed.com) |
+| ky | 1.14.3 | MIT | Copyright (c) Sindre Sorhus (https://sindresorhus.com) |
+| level-supports | 6.2.0 | MIT | Copyright (c) 2019 The contributors to level-supports. |
+| level-transcoder | 1.0.1 | MIT | Copyright © 2012 The contributors to level-transcoder and level-codec. |
+| libnpmaccess | 10.0.3 | ISC | Copyright npm, Inc |
+| libnpmdiff | 8.1.6 | ISC | Copyright (c) GitHub Inc. |
+| libnpmfund | 7.0.20 | ISC | Copyright (c) npm Inc. |
+| libnpmorg | 8.0.1 | ISC | Copyright npm, Inc |
+| libnpmpack | 9.1.6 | ISC | Copyright npm, Inc |
+| libnpmpublish | 11.1.3 | ISC | Copyright npm, Inc |
+| libnpmsearch | 9.0.1 | ISC | Copyright npm, Inc |
+| libnpmteam | 8.0.2 | ISC | Copyright npm, Inc |
+| libnpmversion | 8.0.3 | ISC | Copyright (c) Isaac Z. Schlueter |
+| lodash | 4.18.1 | MIT | Copyright OpenJS Foundation and other contributors <https://openjsf.org/> |
+| lodash.camelcase | 4.3.0 | MIT | Copyright jQuery Foundation and other contributors <https://jquery.org/> |
+| lodash.debounce | 4.0.8 | MIT | Copyright jQuery Foundation and other contributors <https://jquery.org/> |
+| lodash.includes | 4.3.0 | MIT | Copyright jQuery Foundation and other contributors <https://jquery.org/> |
+| lodash.isboolean | 3.0.3 | MIT | Copyright 2012-2016 The Dojo Foundation <http://dojofoundation.org/> |
+| lodash.isinteger | 4.0.4 | MIT | Copyright jQuery Foundation and other contributors <https://jquery.org/> |
+| lodash.isnumber | 3.0.3 | MIT | Copyright 2012-2016 The Dojo Foundation <http://dojofoundation.org/> |
+| lodash.isplainobject | 4.0.6 | MIT | Copyright jQuery Foundation and other contributors <https://jquery.org/> |
+| lodash.isstring | 4.0.1 | MIT | Copyright 2012-2016 The Dojo Foundation <http://dojofoundation.org/> |
+| lodash.once | 4.1.1 | MIT | Copyright jQuery Foundation and other contributors <https://jquery.org/> |
+| log-symbols | 4.1.0 | MIT | Copyright (c) Sindre Sorhus (https://sindresorhus.com) |
+| long | 5.3.2 | Apache-2.0 | - |
+| lower-case | 2.0.2 | MIT | Copyright (c) 2014 Blake Embrey |
+| lru-cache | 10.4.3 | ISC | Copyright (c) 2010-2023 Isaac Z. Schlueter and Contributors |
+| make-dir | 2.1.0 | MIT | Copyright (c) Sindre Sorhus (sindresorhus.com) |
+| make-fetch-happen | 15.0.5 | ISC | Copyright 2017-2022 (c) npm, Inc. |
+| math-intrinsics | 1.1.0 | MIT | Copyright (c) 2024 ECMAScript Shims |
+| maybe-combine-errors | 1.0.0 | MIT | - |
+| md5 | 2.3.0 | BSD-3-Clause | Copyright © 2011-2012, Paul Vorbach. |
+| merge-stream | 2.0.0 | MIT | Copyright (c) Stephen Sugden (stephensugden.com) |
+| mime | 1.6.0 | MIT | Copyright (c) 2010 Benjamin Thomas, Robert Kieffer |
+| mime-types | 2.1.35 | MIT | Copyright (c) 2014 Jonathan Ong |
+| mimic-response | 3.1.0 | MIT | Copyright (c) Sindre Sorhus (https://sindresorhus.com) |
+| minipass | 3.3.6 | ISC | Copyright (c) 2017-2022 npm, Inc., Isaac Z. Schlueter, and Contributors |
+| minipass-collect | 2.0.1 | ISC | Copyright (c) 2019-2023 Isaac Z. Schlueter and Contributors |
+| minipass-pipeline | 1.2.4 | ISC | Copyright (c) Isaac Z. Schlueter and Contributors |
+| minipass-sized | 2.0.0 | ISC | Copyright (c) Isaac Z. Schlueter and Contributors |
+| minizlib | 3.1.0 | MIT | Copyright (c) 2017-2023 Isaac Z. Schlueter and Contributors |
+| mkdirp | 0.5.6 | MIT | Copyright 2010 James Halliday |
+| module-error | 1.0.2 | MIT | Copyright (c) Vincent Weevers |
+| moo | 0.5.3 | BSD-3-Clause | Copyright (c) 2017, Tim Radvan (tjvr) |
+| ms | 2.1.3 | MIT | Copyright (c) 2020 Vercel, Inc. |
+| mute-stream | 3.0.0 | ISC | Copyright (c) Isaac Z. Schlueter and Contributors |
+| napi-build-utils | 2.0.0 | MIT | Copyright (c) 2018 inspiredware |
+| negotiator | 1.0.0 | MIT | Copyright (c) 2012-2014 Federico Romero |
+| next-tick | 1.0.0 | MIT | Copyright (C) 2012-2016 Mariusz Nowak |
+| no-case | 3.0.4 | MIT | Copyright (c) 2014 Blake Embrey |
+| node-abi | 3.92.0 | MIT | Copyright (c) 2016 Lukas Geiger |
+| node-addon-api | 5.1.0 | MIT | Copyright (c) 2017 Node.js API collaborators |
+| node-fetch | 2.7.0 | MIT | Copyright (c) 2016 David Frank |
+| node-gyp | 12.3.0 | MIT | Copyright (c) 2012 Nathan Rajlich |
+| node-pty | 1.1.0-beta39 | MIT | Copyright (c) 2012-2015, Christopher Jeffrey (https://github.com/chjj/) |
+| nopt | 9.0.0 | ISC | Copyright (c) Isaac Z. Schlueter and Contributors |
+| normalize-path | 3.0.0 | MIT | Copyright (c) 2014-2018, Jon Schlinkert. |
+| npm-audit-report | 7.0.0 | ISC | Copyright (c) npm, Inc. |
+| npm-bundled | 5.0.0 | ISC | Copyright (c) npm, Inc. and Contributors |
+| npm-install-checks | 8.0.0 | BSD-2-Clause | Copyright (c) Robert Kowalski and Isaac Z. Schlueter ("Authors") |
+| npm-normalize-package-bin | 5.0.0 | ISC | Copyright (c) npm, Inc. |
+| npm-package-arg | 13.0.2 | ISC | Copyright (c) npm, Inc. |
+| npm-packlist | 10.0.4 | ISC | Copyright (c) Isaac Z. Schlueter and Contributors |
+| npm-pick-manifest | 11.0.3 | ISC | Copyright (c) npm, Inc. |
+| npm-profile | 12.0.1 | ISC | Copyright npm, Inc. |
+| npm-registry-fetch | 19.1.1 | ISC | Copyright npm, Inc. |
+| npm-run-path | 4.0.1 | MIT | Copyright (c) Sindre Sorhus (sindresorhus.com) |
+| npm-user-validate | 4.0.0 | BSD-2-Clause | Copyright (c) Robert Kowalski |
+| ollama-ai-agent | 1.3.5 | MIT | Copyright (c) 2025 Nishant Unavane |
+| on-finished | 2.4.1 | MIT | Copyright (c) 2013 Jonathan Ong |
+| once | 1.4.0 | ISC | Copyright (c) Isaac Z. Schlueter and Contributors |
+| onetime | 5.1.2 | MIT | Copyright (c) Sindre Sorhus (https://sindresorhus.com) |
+| open | 10.2.0 | MIT | Copyright (c) Sindre Sorhus (https://sindresorhus.com) |
+| openid-client | 6.8.4 | MIT | Copyright (c) 2016 Filip Skokan |
+| ora | 5.4.1 | MIT | Copyright (c) Sindre Sorhus (https://sindresorhus.com) |
+| os-tmpdir | 1.0.2 | MIT | Copyright (c) Sindre Sorhus (sindresorhus.com) |
+| p-event | 4.2.0 | MIT | Copyright (c) Sindre Sorhus (https://sindresorhus.com) |
+| p-finally | 1.0.0 | MIT | Copyright (c) Sindre Sorhus (sindresorhus.com) |
+| p-map | 7.0.4 | MIT | Copyright (c) Sindre Sorhus (https://sindresorhus.com) |
+| p-timeout | 3.2.0 | MIT | Copyright (c) Sindre Sorhus (sindresorhus.com) |
+| package-json | 10.0.1 | MIT | Copyright (c) Sindre Sorhus (https://sindresorhus.com) |
+| param-case | 3.0.4 | MIT | Copyright (c) 2014 Blake Embrey |
+| parse-conflict-json | 5.0.1 | ISC | Copyright npm, Inc. |
+| parseurl | 1.3.3 | MIT | Copyright (c) 2014 Jonathan Ong |
+| pascal-case | 3.1.2 | MIT | Copyright (c) 2014 Blake Embrey |
+| path-is-absolute | 1.0.1 | MIT | Copyright (c) Sindre Sorhus (sindresorhus.com) |
+| path-key | 3.1.1 | MIT | Copyright (c) Sindre Sorhus (sindresorhus.com) |
+| path-parse | 1.0.7 | MIT | Copyright (c) 2015 Javier Blanco |
+| pem | 1.14.8 | MIT | Copyright (c) 2012 Andris Reinman |
+| picocolors | 1.1.1 | ISC | Copyright (c) 2021-2024 Oleksii Raspopov, Kostiantyn Denysov, Anton Verinov |
+| picomatch | 4.0.4 | MIT | Copyright (c) 2017-present, Jon Schlinkert. |
+| pify | 4.0.1 | MIT | Copyright (c) Sindre Sorhus (sindresorhus.com) |
+| pixelmatch | 7.1.0 | ISC | Copyright (c) 2025, Mapbox |
+| playwright | 1.57.0 | Apache-2.0 | - |
+| playwright-core | 1.57.0 | Apache-2.0 | - |
+| pngjs | 7.0.0 | MIT | - |
+| prebuild-install | 7.1.3 | MIT | Copyright (c) 2015 Mathias Buus |
+| proc-log | 6.1.0 | ISC | Copyright (c) GitHub, Inc. |
+| process-nextick-args | 2.0.1 | MIT | - |
+| proggy | 4.0.0 | ISC | Copyright (c) GitHub, Inc. |
+| promise-all-reject-late | 1.0.1 | ISC | Copyright (c) Isaac Z. Schlueter |
+| promise-call-limit | 3.0.2 | ISC | Copyright (c) Isaac Z. Schlueter |
+| prompts | 2.4.2 | MIT | Copyright (c) 2018 Terkel Gjervig Nielsen |
+| promzard | 3.0.1 | ISC | Copyright (c) Isaac Z. Schlueter |
+| proto-list | 1.2.4 | ISC | Copyright (c) Isaac Z. Schlueter and Contributors |
+| protobufjs | 7.6.6 | BSD-3-Clause | Copyright (c) 2016, Daniel Wirtz  All rights reserved. |
+| proxy-from-env | 2.1.0 | MIT | Copyright (C) 2016-2018 Rob Wu |
+| punycode | 2.3.1 | MIT | Copyright Mathias Bynens <https://mathiasbynens.be/> |
+| pupa | 3.3.0 | MIT | Copyright (c) Sindre Sorhus (https://sindresorhus.com) |
+| python | 2026.4.0 | MIT | Copyright (c) Microsoft Corporation. All rights reserved. |
+| qrcode-terminal | 0.12.0 | Apache 2.0 | - |
+| randexp | 0.4.6 | MIT | Copyright (C) 2011 by Roly Fentanes |
+| range-parser | 1.2.1 | MIT | Copyright (c) 2012-2014 TJ Holowaychuk |
+| rc | 1.2.8 | (BSD-2-Clause OR MIT OR Apache-2.0) | Copyright (c) 2011 Dominic Tarr |
+| read | 5.0.1 | ISC | Copyright (c) Isaac Z. Schlueter and Contributors |
+| read-cmd-shim | 6.0.0 | ISC | Copyright (c) 2015, Rebecca Turner |
+| readdirp | 3.6.0 | MIT | Copyright (c) 2012-2019 Thorsten Lorenz, Paul Miller (https://paulmillr.com) |
+| regenerate | 1.4.2 | MIT | Copyright Mathias Bynens <https://mathiasbynens.be/> |
+| regenerate-unicode-properties | 10.1.1 | MIT | Copyright Mathias Bynens <https://mathiasbynens.be/> |
+| regenerator-transform | 0.15.2 | MIT | Copyright (c) 2014-present, Facebook, Inc. |
+| regexpu-core | 5.3.2 | MIT | Copyright Mathias Bynens <https://mathiasbynens.be/> |
+| registry-auth-token | 5.1.1 | MIT | Copyright (c) 2016 Espen Hovlandsdal |
+| relateurl | 0.2.7 | MIT | Copyright (c) Steven Vachon (svachon.com) |
+| remove-trailing-separator | 1.1.0 | ISC | - |
+| replace-ext | 1.0.1 | MIT | Copyright (c) 2014 Blaine Bublitz, Eric Schoffstall and other contributors |
+| require-directory | 2.1.1 | MIT | Copyright (c) 2011 Troy Goode |
+| resolve | 1.22.8 | MIT | Copyright (c) 2012 James Halliday |
+| ret | 0.1.15 | MIT | Copyright (C) 2011 by Roly Fentanes |
+| rimraf | 5.0.10 | ISC | Copyright (c) 2011-2023 Isaac Z. Schlueter and Contributors |
+| ruby-lsp | 0.10.3 | MIT | Copyright (c) 2022-present, Shopify Inc. |
+| run-applescript | 7.1.0 | MIT | Copyright (c) Sindre Sorhus (https://sindresorhus.com) |
+| safe-buffer | 5.2.1 | MIT | Copyright (c) Feross Aboukhadijeh |
+| safer-buffer | 2.1.2 | MIT | Copyright (c) 2018 Nikita Skovoroda |
+| semver | 7.8.5 | ISC | Copyright (c) Isaac Z. Schlueter and Contributors |
+| send | 0.19.2 | MIT | Copyright (c) 2012 TJ Holowaychuk |
+| serve-static | 1.16.3 | MIT | Copyright (c) 2010 Sencha Inc. |
+| setprototypeof | 1.2.0 | ISC | Copyright (c) 2015, Wes Todd |
+| sharp | 0.35.4 | Apache-2.0 | - |
+| shebang-command | 2.0.0 | MIT | Copyright (c) Kevin Mårtensson (github.com/kevva) |
+| signal-exit | 4.1.0 | ISC | Copyright (c) 2015-2023 Benjamin Coe, Isaac Z. Schlueter, and Contributors |
+| sigstore | 4.1.0 | Apache-2.0 | - |
+| simple-concat | 1.0.1 | MIT | Copyright (c) Feross Aboukhadijeh |
+| simple-get | 4.0.1 | MIT | Copyright (c) Feross Aboukhadijeh |
+| sisteransi | 1.0.5 | MIT | Copyright (c) 2018 Terkel Gjervig Nielsen |
+| smart-buffer | 4.2.0 | MIT | Copyright (c) 2013-2017 Josh Glazebrook |
+| smtp-address-parser | 1.0.10 | MIT | Copyright (c) 2021 Gene Hightower |
+| socks | 2.8.7 | MIT | Copyright (c) 2013 Josh Glazebrook |
+| socks-proxy-agent | 8.0.5 | MIT | Copyright (c) 2013 Nathan Rajlich |
+| source-map | 0.6.1 | BSD-3-Clause | Copyright (c) 2009-2011, Mozilla Foundation and contributors |
+| spawn-command | 0.0.2 | MIT | Copyright (C) 2011 by Maciej Małecki |
+| spdx-expression-parse | 4.0.0 | MIT | Copyright (c) 2015 Kyle E. Mitchell & other authors listed in AUTHORS |
+| ssri | 13.0.1 | ISC | Copyright 2021 (c) npm, Inc. |
+| statuses | 2.0.2 | MIT | Copyright (c) 2014 Jonathan Ong |
+| steamworks.js | 0.4.0 | MIT | Copyright (c) 2022 Gabriel Francisco Dos Santos |
+| string-width | 7.2.0 | MIT | Copyright (c) Sindre Sorhus (https://sindresorhus.com) |
+| string_decoder | 1.3.0 | MIT | Copyright Node.js contributors. All rights reserved. |
+| strip-ansi | 7.2.0 | MIT | Copyright (c) Sindre Sorhus (https://sindresorhus.com) |
+| strip-final-newline | 2.0.0 | MIT | Copyright (c) Sindre Sorhus (sindresorhus.com) |
+| strip-json-comments | 2.0.1 | MIT | Copyright (c) Sindre Sorhus (sindresorhus.com) |
+| stubborn-fs | 2.0.0 | MIT | Copyright (c) 2022-present Fabio Spampinato |
+| stubborn-utils | 1.0.2 | MIT | Copyright (c) 2025-present Fabio Spampinato |
+| sudo-prompt | 8.2.5 | MIT | Copyright (c) 2015 Joran Dirk Greef |
+| supports-color | 10.2.2 | MIT | Copyright (c) Sindre Sorhus (https://sindresorhus.com) |
+| supports-preserve-symlinks-flag | 1.0.0 | MIT | Copyright (c) 2022 Inspect JS |
+| tar-fs | 2.1.4 | MIT | Copyright (c) 2014 Mathias Buus |
+| terser | 5.44.1 | BSD-2-Clause | Copyright 2012-2018 (c) Mihai Bazon |
+| tiny-relative-date | 2.0.2 | MIT | Copyright (c) 2017 Joseph Wynn |
+| tinyglobby | 0.2.16 | MIT | Copyright (c) 2024 Madeline Gurriarán |
+| tmp | 0.0.33 | MIT | Copyright (c) 2014 KARASZI István |
+| to-regex-range | 5.0.1 | MIT | Copyright (c) 2015-present, Jon Schlinkert. |
+| toidentifier | 1.0.1 | MIT | Copyright (c) 2016 Douglas Christopher Wilson |
+| tree-kill | 1.2.2 | MIT | Copyright (c) 2018 Peter Krumins |
+| treeverse | 3.0.0 | ISC | Copyright (c) npm, Inc. and Contributors |
+| tuf-js | 4.1.0 | MIT | Copyright (c) 2022 GitHub and the TUF Contributors |
+| tunnel-agent | 0.6.0 | Apache-2.0 | - |
+| typedarray-to-buffer | 3.1.5 | MIT | Copyright (c) Feross Aboukhadijeh |
+| typescript | 6.0.3 | Apache-2.0 | - |
+| underscore | 1.13.8 | MIT | Copyright (c) 2009-2022 Jeremy Ashkenas, Julian Gonggrijp, and DocumentCloud and Investigative Reporters & Editors |
+| unicode-canonical-property-names-ecmascript | 2.0.0 | MIT | Copyright Mathias Bynens <https://mathiasbynens.be/> |
+| unicode-match-property-ecmascript | 2.0.0 | MIT | Copyright Mathias Bynens <https://mathiasbynens.be/> |
+| unicode-match-property-value-ecmascript | 2.1.0 | MIT | Copyright Mathias Bynens <https://mathiasbynens.be/> |
+| unpipe | 1.0.0 | MIT | Copyright (c) 2015 Douglas Christopher Wilson |
+| update-browserslist-db | 1.1.4 | MIT | Copyright 2022 Andrey Sitnik and other contributors |
+| update-notifier | 7.3.1 | BSD-2-Clause | Copyright Google |
+| util-deprecate | 1.0.2 | MIT | Copyright (c) 2014 Nathan Rajlich |
+| uuid | 11.1.1 | MIT | Copyright (c) 2010-2020 Robert Kieffer and other contributors |
+| validate-npm-package-name | 7.0.2 | ISC | Copyright (c) 2015, npm, Inc |
+| vinyl | 2.2.1 | MIT | Copyright (c) 2013 Blaine Bublitz, Eric Schoffstall and other contributors |
+| vinyl-sourcemaps-apply | 0.2.1 | ISC | - |
+| vscode-python-envs | 1.20.1 | MIT | Copyright (c) Microsoft Corporation. |
+| wait-on | 7.2.0 | MIT | Copyright (c) 2015 Jeff Barczewski |
+| walk-up-path | 4.0.0 | ISC | Copyright (c) 2020-2023 Isaac Z. Schlueter |
+| wcwidth | 1.0.1 | MIT | Copyright (C) 2012 by Jun Woong. |
+| webidl-conversions | 3.0.1 | BSD-2-Clause | Copyright (c) 2014, Domenic Denicola |
+| websocket | 1.0.32 | Apache-2.0 | - |
+| whatwg-url | 5.0.0 | MIT | Copyright (c) 2015–2016 Sebastian Mayr |
+| which | 6.0.1 | ISC | Copyright (c) Isaac Z. Schlueter and Contributors |
+| windows-notification-state | 2.0.0 | MIT | Copyright (c) 2017 Felix Rieseberg |
+| wordwrap | 1.0.0 | MIT | - |
+| workerpool | 9.3.3 | Apache-2.0 | - |
+| wrap-ansi | 9.0.2 | MIT | Copyright (c) Sindre Sorhus (https://sindresorhus.com) |
+| wrappy | 1.0.2 | ISC | Copyright (c) Isaac Z. Schlueter and Contributors |
+| write-file-atomic | 7.0.1 | ISC | Copyright (c) 2015, Rebecca Turner |
+| ws | 8.18.3 | MIT | Copyright (c) 2011 Einar Otto Stangvik |
+| wsl-utils | 0.1.0 | MIT | Copyright (c) Sindre Sorhus (https://sindresorhus.com) |
+| xdg-basedir | 5.1.0 | MIT | Copyright (c) Sindre Sorhus (https://sindresorhus.com) |
+| y18n | 5.0.8 | ISC | Copyright (c) 2015, Contributors |
+| yaeti | 0.0.6 | MIT | Copyright (c) 2015 Iñaki Baz Castillo, |
+| yallist | 4.0.0 | ISC | Copyright (c) Isaac Z. Schlueter and Contributors |
+| yargs-parser | 21.1.1 | ISC | Copyright (c) 2016, Contributors |
+| zod | 4.4.3 | MIT | Copyright (c) 2025 Colin McDonnell |

config.json

@@ -0,0 +1,54 @@
+{
+  "architectures": [
+    "Qwen2ForCausalLM"
+  ],
+  "attention_dropout": 0.0,
+  "bos_token_id": 151643,
+  "dtype": "bfloat16",
+  "eos_token_id": 151643,
+  "hidden_act": "silu",
+  "hidden_size": 896,
+  "initializer_range": 0.02,
+  "intermediate_size": 4864,
+  "layer_types": [
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention"
+  ],
+  "max_position_embeddings": 32768,
+  "max_window_layers": 24,
+  "model_type": "qwen2",
+  "num_attention_heads": 14,
+  "num_hidden_layers": 24,
+  "num_key_value_heads": 2,
+  "rms_norm_eps": 1e-06,
+  "rope_scaling": null,
+  "rope_theta": 1000000.0,
+  "sliding_window": null,
+  "tie_word_embeddings": true,
+  "transformers_version": "4.57.6",
+  "use_cache": false,
+  "use_sliding_window": false,
+  "vocab_size": 151936
+}

evaluation/val_v2c_summary.json

@@ -0,0 +1,640 @@
+{
+  "schema": "kapi-hf-eval-summary/1",
+  "kind": "val_v2c",
+  "label": "Internal evaluation on synthetic data; no blind human-written set. Aggregates only; no item text.",
+  "model": "esilva/KapiCoder-Mongo-Nano",
+  "split": "val",
+  "n": 2000,
+  "seed": 20261008,
+  "contract": "editor-context-v1",
+  "dataset": {
+    "sample_ids_sha256": "28ba86be496839a321de0035d6a7e21827a5e8de4362151d80958ee8c798562f",
+    "sha256": "af3331481966f3ebfb05d36109b43014cd850f669f6f5d2a524f8da6e9609ec7",
+    "manifest_sha256": "2fef4f342d8a21eaf6d7e9eb1f223b3c8094d56c8dcd00e64f04bc20b20cd3cb",
+    "stratify_by": [
+      "cut_kind",
+      "domain"
+    ],
+    "cut_kind": {
+      "multiline": 161,
+      "key": 319,
+      "value": 280,
+      "rest": 241,
+      "operator": 268,
+      "member": 201,
+      "stage": 251,
+      "partial": 200,
+      "stmt_start": 79
+    },
+    "domain": {
+      "commerce": 168,
+      "content": 161,
+      "energy": 171,
+      "finance": 175,
+      "gaming": 163,
+      "healthcare-synthetic": 173,
+      "hr": 168,
+      "iot": 168,
+      "logistics": 157,
+      "saas": 165,
+      "travel": 162,
+      "workspace": 169
+    }
+  },
+  "decoding": "greedy (argmax), padding do vocab mascarado",
+  "max_new_tokens": 96,
+  "context_tokens": 2048,
+  "load": "bf16",
+  "this_model": {
+    "headline": {
+      "n": 2000,
+      "exact_match": 0.405,
+      "normalized_match": 0.405,
+      "prefix_match": 0.4105,
+      "edit_similarity": 0.6499,
+      "chars_accepted": 7.9775,
+      "tab_chars": 6.343,
+      "accepted_prefix_tokens": 2.62,
+      "syntax_valid": 0.9955,
+      "mongo_valid": 0.992,
+      "exec_valid": 0.5945,
+      "exec_valid_incl_empty": 0.5945,
+      "nonexistent_fields_rate": 0.0265,
+      "empty_rate": 0.0,
+      "eos_or_suffix_stop_rate": 0.997,
+      "mean_generated_tokens": 15.292
+    },
+    "headline_ide32": {
+      "n": 2000,
+      "exact_match": 0.4035,
+      "normalized_match": 0.4035,
+      "prefix_match": 0.4105,
+      "edit_similarity": 0.6421,
+      "chars_accepted": 7.9575,
+      "tab_chars": 6.323,
+      "accepted_prefix_tokens": 2.6125,
+      "syntax_valid": 0.836,
+      "mongo_valid": 0.833,
+      "exec_valid": 0.5055,
+      "exec_valid_incl_empty": 0.5055,
+      "nonexistent_fields_rate": 0.0165,
+      "empty_rate": 0.0,
+      "eos_or_suffix_stop_rate": null,
+      "mean_generated_tokens": null
+    },
+    "reference_ceiling": {
+      "syntax_valid": 1.0,
+      "mongo_valid": 1.0,
+      "exec_valid": 0.6255,
+      "nonexistent_fields_rate": 0.0
+    },
+    "exec_stage_distribution": {
+      "ok": 1189,
+      "api": 483,
+      "syntax": 248,
+      "fields": 75,
+      "exec": 2,
+      "type": 3
+    },
+    "by_cut_kind": {
+      "key": {
+        "n": 319,
+        "exact_match": 0.4639,
+        "normalized_match": 0.4639,
+        "prefix_match": 0.4639,
+        "edit_similarity": 0.6428,
+        "chars_accepted": 6.2351,
+        "tab_chars": 5.7335,
+        "accepted_prefix_tokens": 1.9154,
+        "syntax_valid": 0.9969,
+        "mongo_valid": 0.9937,
+        "exec_valid": 0.5392,
+        "exec_valid_incl_empty": 0.5392,
+        "nonexistent_fields_rate": 0.0376,
+        "empty_rate": 0.0,
+        "eos_or_suffix_stop_rate": 1.0,
+        "mean_generated_tokens": 8.7147
+      },
+      "member": {
+        "n": 201,
+        "exact_match": 0.7711,
+        "normalized_match": 0.7711,
+        "prefix_match": 0.7761,
+        "edit_similarity": 0.8523,
+        "chars_accepted": 7.9254,
+        "tab_chars": 7.6368,
+        "accepted_prefix_tokens": 1.6269,
+        "syntax_valid": 1.0,
+        "mongo_valid": 0.995,
+        "exec_valid": 0.6368,
+        "exec_valid_incl_empty": 0.6368,
+        "nonexistent_fields_rate": 0.0,
+        "empty_rate": 0.0,
+        "eos_or_suffix_stop_rate": 1.0,
+        "mean_generated_tokens": 5.209
+      },
+      "multiline": {
+        "n": 161,
+        "exact_match": 0.0,
+        "normalized_match": 0.0,
+        "prefix_match": 0.0,
+        "edit_similarity": 0.4178,
+        "chars_accepted": 11.5155,
+        "tab_chars": 7.1491,
+        "accepted_prefix_tokens": 3.5839,
+        "syntax_valid": 0.9938,
+        "mongo_valid": 0.9938,
+        "exec_valid": 0.5901,
+        "exec_valid_incl_empty": 0.5901,
+        "nonexistent_fields_rate": 0.087,
+        "empty_rate": 0.0,
+        "eos_or_suffix_stop_rate": 0.9876,
+        "mean_generated_tokens": 43.7578
+      },
+      "operator": {
+        "n": 268,
+        "exact_match": 0.6045,
+        "normalized_match": 0.6045,
+        "prefix_match": 0.6157,
+        "edit_similarity": 0.7894,
+        "chars_accepted": 7.6194,
+        "tab_chars": 6.4701,
+        "accepted_prefix_tokens": 3.0261,
+        "syntax_valid": 0.9963,
+        "mongo_valid": 0.9963,
+        "exec_valid": 0.6306,
+        "exec_valid_incl_empty": 0.6306,
+        "nonexistent_fields_rate": 0.0187,
+        "empty_rate": 0.0,
+        "eos_or_suffix_stop_rate": 0.9963,
+        "mean_generated_tokens": 9.8881
+      },
+      "partial": {
+        "n": 200,
+        "exact_match": 0.695,
+        "normalized_match": 0.695,
+        "prefix_match": 0.695,
+        "edit_similarity": 0.8264,
+        "chars_accepted": 6.34,
+        "tab_chars": 5.425,
+        "accepted_prefix_tokens": 2.405,
+        "syntax_valid": 0.995,
+        "mongo_valid": 0.98,
+        "exec_valid": 0.57,
+        "exec_valid_incl_empty": 0.57,
+        "nonexistent_fields_rate": 0.05,
+        "empty_rate": 0.0,
+        "eos_or_suffix_stop_rate": 1.0,
+        "mean_generated_tokens": 7.9
+      },
+      "rest": {
+        "n": 241,
+        "exact_match": 0.3444,
+        "normalized_match": 0.3444,
+        "prefix_match": 0.3485,
+        "edit_similarity": 0.6293,
+        "chars_accepted": 9.4855,
+        "tab_chars": 7.3361,
+        "accepted_prefix_tokens": 2.9793,
+        "syntax_valid": 0.9959,
+        "mongo_valid": 0.9876,
+        "exec_valid": 0.5809,
+        "exec_valid_incl_empty": 0.5809,
+        "nonexistent_fields_rate": 0.0124,
+        "empty_rate": 0.0,
+        "eos_or_suffix_stop_rate": 1.0,
+        "mean_generated_tokens": 16.7054
+      },
+      "stage": {
+        "n": 251,
+        "exact_match": 0.012,
+        "normalized_match": 0.012,
+        "prefix_match": 0.012,
+        "edit_similarity": 0.4255,
+        "chars_accepted": 11.4861,
+        "tab_chars": 7.8247,
+        "accepted_prefix_tokens": 4.243,
+        "syntax_valid": 0.992,
+        "mongo_valid": 0.992,
+        "exec_valid": 0.5817,
+        "exec_valid_incl_empty": 0.5817,
+        "nonexistent_fields_rate": 0.0239,
+        "empty_rate": 0.0,
+        "eos_or_suffix_stop_rate": 0.992,
+        "mean_generated_tokens": 30.7131
+      },
+      "stmt_start": {
+        "n": 79,
+        "exact_match": 0.3671,
+        "normalized_match": 0.3671,
+        "prefix_match": 0.3671,
+        "edit_similarity": 0.6448,
+        "chars_accepted": 16.0633,
+        "tab_chars": 13.8734,
+        "accepted_prefix_tokens": 3.5316,
+        "syntax_valid": 0.9873,
+        "mongo_valid": 0.9873,
+        "exec_valid": 0.7468,
+        "exec_valid_incl_empty": 0.7468,
+        "nonexistent_fields_rate": 0.038,
+        "empty_rate": 0.0,
+        "eos_or_suffix_stop_rate": 0.9873,
+        "mean_generated_tokens": 16.3924
+      },
+      "value": {
+        "n": 280,
+        "exact_match": 0.325,
+        "normalized_match": 0.325,
+        "prefix_match": 0.3464,
+        "edit_similarity": 0.6071,
+        "chars_accepted": 2.7536,
+        "tab_chars": 1.8714,
+        "accepted_prefix_tokens": 1.325,
+        "syntax_valid": 0.9964,
+        "mongo_valid": 0.9964,
+        "exec_valid": 0.5929,
+        "exec_valid_incl_empty": 0.5929,
+        "nonexistent_fields_rate": 0.0,
+        "empty_rate": 0.0,
+        "eos_or_suffix_stop_rate": 1.0,
+        "mean_generated_tokens": 8.7571
+      }
+    },
+    "by_domain": {
+      "commerce": {
+        "n": 168,
+        "exact_match": 0.375,
+        "normalized_match": 0.375,
+        "prefix_match": 0.381,
+        "edit_similarity": 0.623,
+        "chars_accepted": 6.9881,
+        "tab_chars": 5.3214,
+        "accepted_prefix_tokens": 2.25,
+        "syntax_valid": 1.0,
+        "mongo_valid": 0.994,
+        "exec_valid": 0.6548,
+        "exec_valid_incl_empty": 0.6548,
+        "nonexistent_fields_rate": 0.0298,
+        "empty_rate": 0.0,
+        "eos_or_suffix_stop_rate": 1.0,
+        "mean_generated_tokens": 14.1786
+      },
+      "content": {
+        "n": 161,
+        "exact_match": 0.3913,
+        "normalized_match": 0.3913,
+        "prefix_match": 0.4099,
+        "edit_similarity": 0.6281,
+        "chars_accepted": 8.0497,
+        "tab_chars": 6.5652,
+        "accepted_prefix_tokens": 2.7019,
+        "syntax_valid": 1.0,
+        "mongo_valid": 1.0,
+        "exec_valid": 0.5901,
+        "exec_valid_incl_empty": 0.5901,
+        "nonexistent_fields_rate": 0.0062,
+        "empty_rate": 0.0,
+        "eos_or_suffix_stop_rate": 1.0,
+        "mean_generated_tokens": 14.8012
+      },
+      "energy": {
+        "n": 171,
+        "exact_match": 0.4269,
+        "normalized_match": 0.4269,
+        "prefix_match": 0.4327,
+        "edit_similarity": 0.6494,
+        "chars_accepted": 7.7953,
+        "tab_chars": 6.5146,
+        "accepted_prefix_tokens": 2.3509,
+        "syntax_valid": 0.9942,
+        "mongo_valid": 0.9942,
+        "exec_valid": 0.614,
+        "exec_valid_incl_empty": 0.614,
+        "nonexistent_fields_rate": 0.0351,
+        "empty_rate": 0.0,
+        "eos_or_suffix_stop_rate": 0.9942,
+        "mean_generated_tokens": 16.6842
+      },
+      "finance": {
+        "n": 175,
+        "exact_match": 0.3486,
+        "normalized_match": 0.3486,
+        "prefix_match": 0.3543,
+        "edit_similarity": 0.6135,
+        "chars_accepted": 8.1257,
+        "tab_chars": 6.6286,
+        "accepted_prefix_tokens": 2.88,
+        "syntax_valid": 0.9943,
+        "mongo_valid": 0.9943,
+        "exec_valid": 0.5943,
+        "exec_valid_incl_empty": 0.5943,
+        "nonexistent_fields_rate": 0.04,
+        "empty_rate": 0.0,
+        "eos_or_suffix_stop_rate": 1.0,
+        "mean_generated_tokens": 18.5257
+      },
+      "gaming": {
+        "n": 163,
+        "exact_match": 0.3681,
+        "normalized_match": 0.3681,
+        "prefix_match": 0.3804,
+        "edit_similarity": 0.6296,
+        "chars_accepted": 7.6196,
+        "tab_chars": 6.0184,
+        "accepted_prefix_tokens": 2.5092,
+        "syntax_valid": 0.9877,
+        "mongo_valid": 0.9755,
+        "exec_valid": 0.5951,
+        "exec_valid_incl_empty": 0.5951,
+        "nonexistent_fields_rate": 0.0245,
+        "empty_rate": 0.0,
+        "eos_or_suffix_stop_rate": 1.0,
+        "mean_generated_tokens": 14.8712
+      },
+      "healthcare-synthetic": {
+        "n": 173,
+        "exact_match": 0.4104,
+        "normalized_match": 0.4104,
+        "prefix_match": 0.4104,
+        "edit_similarity": 0.6618,
+        "chars_accepted": 8.3815,
+        "tab_chars": 6.4624,
+        "accepted_prefix_tokens": 2.7341,
+        "syntax_valid": 0.9942,
+        "mongo_valid": 0.9884,
+        "exec_valid": 0.5607,
+        "exec_valid_incl_empty": 0.5607,
+        "nonexistent_fields_rate": 0.0231,
+        "empty_rate": 0.0,
+        "eos_or_suffix_stop_rate": 0.9942,
+        "mean_generated_tokens": 15.0636
+      },
+      "hr": {
+        "n": 168,
+        "exact_match": 0.4643,
+        "normalized_match": 0.4643,
+        "prefix_match": 0.4762,
+        "edit_similarity": 0.6899,
+        "chars_accepted": 7.8155,
+        "tab_chars": 6.3393,
+        "accepted_prefix_tokens": 2.5357,
+        "syntax_valid": 0.994,
+        "mongo_valid": 0.994,
+        "exec_valid": 0.5833,
+        "exec_valid_incl_empty": 0.5833,
+        "nonexistent_fields_rate": 0.0119,
+        "empty_rate": 0.0,
+        "eos_or_suffix_stop_rate": 0.994,
+        "mean_generated_tokens": 12.7738
+      },
+      "iot": {
+        "n": 168,
+        "exact_match": 0.4107,
+        "normalized_match": 0.4107,
+        "prefix_match": 0.4107,
+        "edit_similarity": 0.6606,
+        "chars_accepted": 8.4821,
+        "tab_chars": 6.4643,
+        "accepted_prefix_tokens": 2.869,
+        "syntax_valid": 1.0,
+        "mongo_valid": 1.0,
+        "exec_valid": 0.6369,
+        "exec_valid_incl_empty": 0.6369,
+        "nonexistent_fields_rate": 0.0536,
+        "empty_rate": 0.0,
+        "eos_or_suffix_stop_rate": 1.0,
+        "mean_generated_tokens": 15.3155
+      },
+      "logistics": {
+        "n": 157,
+        "exact_match": 0.4076,
+        "normalized_match": 0.4076,
+        "prefix_match": 0.4076,
+        "edit_similarity": 0.6721,
+        "chars_accepted": 8.9554,
+        "tab_chars": 6.8408,
+        "accepted_prefix_tokens": 2.8535,
+        "syntax_valid": 1.0,
+        "mongo_valid": 0.9936,
+        "exec_valid": 0.5987,
+        "exec_valid_incl_empty": 0.5987,
+        "nonexistent_fields_rate": 0.0255,
+        "empty_rate": 0.0,
+        "eos_or_suffix_stop_rate": 1.0,
+        "mean_generated_tokens": 15.2102
+      },
+      "saas": {
+        "n": 165,
+        "exact_match": 0.4788,
+        "normalized_match": 0.4788,
+        "prefix_match": 0.4788,
+        "edit_similarity": 0.7021,
+        "chars_accepted": 7.0242,
+        "tab_chars": 5.5636,
+        "accepted_prefix_tokens": 2.3758,
+        "syntax_valid": 0.9939,
+        "mongo_valid": 0.9939,
+        "exec_valid": 0.4848,
+        "exec_valid_incl_empty": 0.4848,
+        "nonexistent_fields_rate": 0.0364,
+        "empty_rate": 0.0,
+        "eos_or_suffix_stop_rate": 1.0,
+        "mean_generated_tokens": 12.4909
+      },
+      "travel": {
+        "n": 162,
+        "exact_match": 0.4136,
+        "normalized_match": 0.4136,
+        "prefix_match": 0.4198,
+        "edit_similarity": 0.6412,
+        "chars_accepted": 8.4506,
+        "tab_chars": 6.6667,
+        "accepted_prefix_tokens": 2.7901,
+        "syntax_valid": 1.0,
+        "mongo_valid": 1.0,
+        "exec_valid": 0.6543,
+        "exec_valid_incl_empty": 0.6543,
+        "nonexistent_fields_rate": 0.0123,
+        "empty_rate": 0.0,
+        "eos_or_suffix_stop_rate": 1.0,
+        "mean_generated_tokens": 16.5123
+      },
+      "workspace": {
+        "n": 169,
+        "exact_match": 0.3669,
+        "normalized_match": 0.3669,
+        "prefix_match": 0.3669,
+        "edit_similarity": 0.6296,
+        "chars_accepted": 8.0828,
+        "tab_chars": 6.7396,
+        "accepted_prefix_tokens": 2.5976,
+        "syntax_valid": 0.9882,
+        "mongo_valid": 0.9763,
+        "exec_valid": 0.568,
+        "exec_valid_incl_empty": 0.568,
+        "nonexistent_fields_rate": 0.0178,
+        "empty_rate": 0.0,
+        "eos_or_suffix_stop_rate": 0.9822,
+        "mean_generated_tokens": 16.8698
+      }
+    },
+    "by_context_kind": {
+      "full": {
+        "n": 1647,
+        "exact_match": 0.4141,
+        "normalized_match": 0.4141,
+        "prefix_match": 0.4196,
+        "edit_similarity": 0.6598,
+        "chars_accepted": 8.1488,
+        "tab_chars": 6.4584,
+        "accepted_prefix_tokens": 2.6752,
+        "syntax_valid": 0.9951,
+        "mongo_valid": 0.9927,
+        "exec_valid": 0.6169,
+        "exec_valid_incl_empty": 0.6169,
+        "nonexistent_fields_rate": 0.0134,
+        "empty_rate": 0.0,
+        "eos_or_suffix_stop_rate": 0.997,
+        "mean_generated_tokens": 15.4244
+      },
+      "irrelevant": {
+        "n": 198,
+        "exact_match": 0.3687,
+        "normalized_match": 0.3687,
+        "prefix_match": 0.3737,
+        "edit_similarity": 0.5935,
+        "chars_accepted": 7.0707,
+        "tab_chars": 5.8081,
+        "accepted_prefix_tokens": 2.2374,
+        "syntax_valid": 1.0,
+        "mongo_valid": 0.9949,
+        "exec_valid": 0.5051,
+        "exec_valid_incl_empty": 0.5051,
+        "nonexistent_fields_rate": 0.0859,
+        "empty_rate": 0.0,
+        "eos_or_suffix_stop_rate": 0.9949,
+        "mean_generated_tokens": 14.5505
+      },
+      "none": {
+        "n": 155,
+        "exact_match": 0.3548,
+        "normalized_match": 0.3548,
+        "prefix_match": 0.3613,
+        "edit_similarity": 0.6167,
+        "chars_accepted": 7.3161,
+        "tab_chars": 5.8,
+        "accepted_prefix_tokens": 2.5226,
+        "syntax_valid": 0.9935,
+        "mongo_valid": 0.9806,
+        "exec_valid": 0.471,
+        "exec_valid_incl_empty": 0.471,
+        "nonexistent_fields_rate": 0.0903,
+        "empty_rate": 0.0,
+        "eos_or_suffix_stop_rate": 1.0,
+        "mean_generated_tokens": 14.8323
+      }
+    }
+  },
+  "metrics_by_model": {
+    "this_model_v3": {
+      "syntax_valid": 0.9955,
+      "mongodb_valid": 0.992,
+      "nonexistent_fields_rate": 2.65,
+      "conversational_suggestion_rate": 0.0,
+      "fim_acc": 0.405,
+      "exact_match": 0.405,
+      "apt_mean": 2.62,
+      "exec_valid_autocomplete": 0.5945,
+      "empty_rate": 0.0,
+      "reference_ceiling_exec_valid": 0.6255,
+      "exec_valid_vs_ceiling": 0.9504,
+      "nonexistent_fields_full_rate_pct": 1.34,
+      "n_context_full": 1647,
+      "exec_valid_full": 0.6169,
+      "nonexistent_fields_irrelevant_rate_pct": 8.59,
+      "n_context_irrelevant": 198,
+      "exec_valid_irrelevant": 0.5051,
+      "nonexistent_fields_none_rate_pct": 9.03,
+      "n_context_none": 155,
+      "exec_valid_none": 0.471,
+      "info_fim_acc_gain_vs_zero_shot_base_pp": 32.9,
+      "info_apt_delta_vs_zero_shot_base": 1.4155,
+      "info_exec_valid_delta_vs_zero_shot_base_pp": 38.15,
+      "generic_js_fim_acc": 0.53,
+      "generic_js_fim_delta_pp_vs_zero_shot_base": 19.0,
+      "write_or_destructive_executions": 0,
+      "train_eval_leak_count": 0,
+      "decontamination_approved": true,
+      "auto_report_attached": true
+    },
+    "previous_internal_checkpoint_v1_not_released": {
+      "syntax_valid": 0.995,
+      "mongodb_valid": 0.994,
+      "nonexistent_fields_rate": 3.8,
+      "conversational_suggestion_rate": 0.0,
+      "fim_acc": 0.403,
+      "exact_match": 0.403,
+      "apt_mean": 2.6605,
+      "exec_valid_autocomplete": 0.582,
+      "empty_rate": 0.0,
+      "reference_ceiling_exec_valid": 0.6255,
+      "exec_valid_vs_ceiling": 0.9305,
+      "nonexistent_fields_full_rate_pct": 2.06,
+      "n_context_full": 1647,
+      "exec_valid_full": 0.6084,
+      "nonexistent_fields_irrelevant_rate_pct": 11.11,
+      "n_context_irrelevant": 198,
+      "exec_valid_irrelevant": 0.4798,
+      "nonexistent_fields_none_rate_pct": 12.9,
+      "n_context_none": 155,
+      "exec_valid_none": 0.4323,
+      "info_fim_acc_gain_vs_zero_shot_base_pp": 32.7,
+      "info_apt_delta_vs_zero_shot_base": 1.456,
+      "info_exec_valid_delta_vs_zero_shot_base_pp": 36.9,
+      "generic_js_fim_acc": 0.518,
+      "generic_js_fim_delta_pp_vs_zero_shot_base": 17.8,
+      "write_or_destructive_executions": 0,
+      "train_eval_leak_count": 0,
+      "decontamination_approved": true,
+      "auto_report_attached": true
+    },
+    "zero_shot_base": {
+      "syntax_valid": 0.498,
+      "mongodb_valid": 0.4275,
+      "nonexistent_fields_rate": 4.85,
+      "conversational_suggestion_rate": 0.0,
+      "fim_acc": 0.076,
+      "exact_match": 0.076,
+      "apt_mean": 1.2045,
+      "exec_valid_autocomplete": 0.213,
+      "empty_rate": 0.0055,
+      "reference_ceiling_exec_valid": 0.6255,
+      "exec_valid_vs_ceiling": 0.3405,
+      "nonexistent_fields_full_rate_pct": 4.13,
+      "n_context_full": 1647,
+      "exec_valid_full": 0.2198,
+      "nonexistent_fields_irrelevant_rate_pct": 8.59,
+      "n_context_irrelevant": 198,
+      "exec_valid_irrelevant": 0.1667,
+      "nonexistent_fields_none_rate_pct": 7.74,
+      "n_context_none": 155,
+      "exec_valid_none": 0.2,
+      "generic_js_fim_acc": 0.34,
+      "write_or_destructive_executions": 0,
+      "train_eval_leak_count": 0,
+      "decontamination_approved": true,
+      "auto_report_attached": true
+    }
+  },
+  "metric_definitions": {
+    "exact_match": "pred.strip() == middle.strip()",
+    "normalized_match": "igualdade com espacos normalizados (FIM accuracy do bake-off)",
+    "chars_accepted": "tamanho do prefixo comum (caracteres) entre predicao e middle",
+    "accepted_prefix_tokens": "tokens iniciais da referencia cobertos pelo prefixo comum (APT)",
+    "syntax_valid": "node vm.Script sobre prefix+pred+suffix",
+    "mongo_valid": "mongo_validate (AST, estatico) sobre o documento reconstruido",
+    "exec_valid": "chat_validate.validate_text stage=ok (sintaxe, API do Console, campos, tipos, execucao somente-leitura em kapi_fx_*) E predicao nao vazia; exec_valid_incl_empty conta vazio como valido",
+    "nonexistent_fields_rate": "stage=fields com 'campo/estrutura invalido' (campo ausente do schema) sobre todos os exemplos",
+    "reference_ceiling": "mesmas checagens aplicadas ao middle de referencia (teto/calibracao do validador)"
+  }
+}

examples/example_fim.py

@@ -0,0 +1,51 @@
+"""Minimal fill-in-the-middle example (transformers, CPU, no IDE required).
+
+    python examples/example_fim.py                 # model files next to this folder (a local clone or snapshot), else the Hub repository
+    python examples/example_fim.py <path-or-repo>  # explicit model directory or repository id
+"""
+import sys
+from pathlib import Path
+
+import torch
+from transformers import AutoModelForCausalLM, AutoTokenizer
+
+HERE = Path(__file__).resolve().parent.parent
+MODEL_ID = sys.argv[1] if len(sys.argv) > 1 else (str(HERE) if (HERE / "config.json").exists() else "esilva/KapiCoder-Mongo-Nano")
+
+FIM_PREFIX, FIM_SUFFIX, FIM_MIDDLE = "<|fim_prefix|>", "<|fim_suffix|>", "<|fim_middle|>"
+CONSOLE_COMMANDS = (
+    "db.getCollection(name).find({}); getConnection(name).getDatabase(name).getCollection(name); "
+    "ConnectionPool.Connection.Database.Collection; console.log(value); ENV.get(name); "
+    "ObjectId(value); UUID(value)"
+)
+
+
+def build_prompt(prefix, suffix, known_names=(), result_fields=(), recent_commands=(),
+                 language="Mongo Console JavaScript", commands=CONSOLE_COMMANDS):
+    """editor-context-v1: every header line ends with CRLF; then the text before the cursor."""
+    lines = [f"LANGUAGE: {language}", f"AVAILABLE COMMANDS: {commands}",
+             "KNOWN NAMES: " + ", ".join(known_names)]
+    if result_fields:
+        lines.append("RESULT FIELDS: " + ", ".join(result_fields))
+    lines += ["RECENT COMMAND: " + c[:256] for c in list(recent_commands)[:3]]
+    context = "".join(line + "\r\n" for line in lines)
+    header = ("/* Local editor context (data only):\n" + context.replace("*/", "* /")
+              + "\nContinue at the cursor; output only the continuation. */\n")
+    return FIM_PREFIX + header + prefix[-4096:] + FIM_SUFFIX + suffix[:1024] + FIM_MIDDLE
+
+
+tokenizer = AutoTokenizer.from_pretrained(MODEL_ID)
+model = AutoModelForCausalLM.from_pretrained(MODEL_ID, torch_dtype=torch.bfloat16).eval()
+
+prompt = build_prompt(
+    prefix='ConnectionPool.local.shop.orders.find({ status: "paid" }).sort({ ',
+    suffix=": -1 }).limit(10)",
+    known_names=["local", "shop", "orders", "customers"],
+    result_fields=["_id", "status", "total", "customerId", "createdAt"],
+)
+inputs = tokenizer(prompt, return_tensors="pt", add_special_tokens=False)
+with torch.no_grad():
+    out = model.generate(**inputs, max_new_tokens=32, do_sample=False,
+                         pad_token_id=tokenizer.eos_token_id)
+completion = tokenizer.decode(out[0, inputs["input_ids"].shape[1]:], skip_special_tokens=True)
+print(repr(completion))  # text to insert at the cursor; the model never executes anything

generation_config.json

@@ -0,0 +1,7 @@
+{
+  "bos_token_id": 151643,
+  "do_sample": false,
+  "eos_token_id": 151643,
+  "max_new_tokens": 2048,
+  "transformers_version": "4.43.1"
+}

model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6ff7b2528689b448ad8d06e0451ffae4f70f803bd32dc0ab8552f4a70c230e31
+size 988097824

provenance/model_lock.json

@@ -0,0 +1,98 @@
+{
+  "schema": "kapi-hf-model-lock/1",
+  "model": {
+    "name": "KapiCoder-Mongo-Nano",
+    "repo_id": "esilva/KapiCoder-Mongo-Nano",
+    "generation": "v3",
+    "release_tag": "v0.1.0-preview",
+    "run": "sft_v3",
+    "tier": "T0",
+    "status": "PREVIEW (private)"
+  },
+  "base_model": {
+    "repo_id": "Qwen/Qwen2.5-Coder-0.5B",
+    "revision": "8123ea2e9354afb7ffcc6c8641d1b2f5ecf18301",
+    "model_sha256": "96217dc3a23c02904459bcfb1dcad72fda4ca5e629182cca700f558803fecbc5",
+    "license": "Apache-2.0"
+  },
+  "weights": {
+    "format": "safetensors",
+    "dtype": "bfloat16",
+    "combined_sha256": "029aab78dde4ee60c0c1a6cb0e5a055b6a81b79ac8224fe64f117a2e04f27e2c",
+    "files_sha256": {
+      "config.json": "1a0023f5853290cf4b94ddb544ef778e8ecf7861f2d3e79e75f90e064863ed9c",
+      "generation_config.json": "215d4506cf90f3590fe821bee47dc92e99caa06f2c1b36076693b5209980009c",
+      "merges.txt": "599bab54075088774b1733fde865d5bd747cbcc7a547c5bc12610e874e26f5e3",
+      "model.safetensors": "6ff7b2528689b448ad8d06e0451ffae4f70f803bd32dc0ab8552f4a70c230e31",
+      "tokenizer.json": "c0382117ea329cdf097041132f6d735924b697924d6f6fc3945713e96ce87539",
+      "tokenizer_config.json": "283ec39d07ebfd1904aa2dee3c7d73524ed96c489fdb3915629900d3ab3ac319",
+      "vocab.json": "ca10d7e9fb3ed18575dd1e277a2579c16d108e32f27439684afa0e10b1440910"
+    }
+  },
+  "training": {
+    "method": "full",
+    "master_dtype": "fp32",
+    "optimizer": "adamw",
+    "peak_lr": 2e-05,
+    "weight_decay": 0.01,
+    "seed": 20261018,
+    "lora": null,
+    "trainable_params": 494032768,
+    "total_steps": 4704,
+    "steps_per_epoch": 2352,
+    "tokens_seen": 129999692,
+    "elapsed_train_h": 4.381,
+    "started_utc": "2026-10-25T01:59:41Z",
+    "finished_utc": "2026-10-25T06:23:50Z",
+    "git_commit": "ba1b880645d86bb33483c016a34782576e901ba9",
+    "git_dirty": false,
+    "train_config_sha256": "8e7ef77567ecdc561c17e6520d952df91bdaa3b33a26964a0979641d09ce68f8",
+    "config_file_sha256": "54fc8718fb07857ded4cf8937c6b785ef4599545cb001924d8cb67cc41b7ecdc",
+    "libraries": {
+      "python": "3.12.10",
+      "torch": "2.10.0+rocm7.14.0a20260611",
+      "transformers": "4.57.6",
+      "peft": "0.17.1",
+      "tokenizers": "0.22.2",
+      "safetensors": "0.7.0"
+    },
+    "hardware": "AMD Radeon RX 7800 XT",
+    "single_training_run": true
+  },
+  "dataset": {
+    "name": "student corpus v3 (not published)",
+    "published": false,
+    "label_source": "mixed: teacher_seq (accepted fim_ctx/completion) | gold (llm, grammar, fim_raw) | replay",
+    "examples": 238453,
+    "tokens_per_epoch": 64999846,
+    "replay_examples": 12618,
+    "replay_token_share": 0.12,
+    "corpus_manifest_sha256": "98abe23cb2b562cb96ee4f0730550f452f08843c2ba6c084a216f347365c2d1d",
+    "ids_sha256": "434f2e2cbe01b80bf9eae45747d91a2660cc152dcbad33f444bef8ea33367cf1",
+    "rows_sha256": "87bcb71bfa0dc4aa6e8e40ab101dd3d5af7ee1ddfbad629bd9180dbc203e1f46",
+    "student_corpus_config_sha256": "47f02a0b6a2a1564d520c4137715de95932835d143003eb7204b01201f2189fb",
+    "accepted_sha256": "2a32311843d3479cfd6ac6210555208bc53d01c19b23887f00a6ab37c1ba97bc"
+  },
+  "evaluation": {
+    "val_v2c": {
+      "split": "val",
+      "n": 2000,
+      "seed": 20261008,
+      "sample_ids_sha256": "28ba86be496839a321de0035d6a7e21827a5e8de4362151d80958ee8c798562f",
+      "dataset_sha256": "af3331481966f3ebfb05d36109b43014cd850f669f6f5d2a524f8da6e9609ec7",
+      "manifest_sha256": "2fef4f342d8a21eaf6d7e9eb1f223b3c8094d56c8dcd00e64f04bc20b20cd3cb"
+    },
+    "homologation_v1": {
+      "set_sha256": "aca0aae51533a95d830c1f296d621f9203a7d76475658fb1e2dcc8f389495332",
+      "n": 6030,
+      "items_published": false
+    },
+    "contract": {
+      "name": "editor-context-v1",
+      "golden_sha256": "06ef5347b128d4dcd11e838ebaa45b30797c5ba30e67222276477e9921ae51c3"
+    }
+  },
+  "publish_tooling": {
+    "scripts_publish_sha256": "c31297cc42c0b967e55d196d2e88b48a073d6dc988cd1d097313faafc6b0d45b"
+  }
+}

provenance/upload_manifest.json

@@ -0,0 +1,84 @@
+{
+  "schema": "kapi-hf-upload-manifest/2",
+  "repo_id": "esilva/KapiCoder-Mongo-Nano",
+  "release_tag": "v0.1.0-preview",
+  "tier": "nano",
+  "note": "sha256 de todos os arquivos do repositorio, exceto este (provenance/upload_manifest.json)",
+  "files": [
+    {
+      "path": "LICENSE",
+      "bytes": 11343,
+      "sha256": "832dd9e00a68dd83b3c3fb9f5588dad7dcf337a0db50f7d9483f310cd292e92e"
+    },
+    {
+      "path": "NOTICE.md",
+      "bytes": 2090,
+      "sha256": "c27b027114ff92cdfa9d0a56cefec91d125e4cb3ca49bae9e4306b2677c1527a"
+    },
+    {
+      "path": "README.md",
+      "bytes": 30918,
+      "sha256": "33878c8d6b14e6a410015aa2405a55c6fbcb93171f643d2dc546ed48e24946ab"
+    },
+    {
+      "path": "REPLAY_ATTRIBUTION.md",
+      "bytes": 47459,
+      "sha256": "0f3f49f1ada3769a05f52c2dcde7de895a766f4605c1422557e963309376f9a6"
+    },
+    {
+      "path": "config.json",
+      "bytes": 1282,
+      "sha256": "1a0023f5853290cf4b94ddb544ef778e8ecf7861f2d3e79e75f90e064863ed9c"
+    },
+    {
+      "path": "evaluation/homologation_v1_summary.json",
+      "bytes": 141103,
+      "sha256": "6caf37deb98617c657a29ba49593160ad7447ea9a09b11633f2b43aeb085fe77"
+    },
+    {
+      "path": "evaluation/val_v2c_summary.json",
+      "bytes": 20613,
+      "sha256": "0f2e30a2eb9117540c17535538fbd80b2ae41cab800f1cc1d02dd583c5dbdc5d"
+    },
+    {
+      "path": "examples/example_fim.py",
+      "bytes": 2642,
+      "sha256": "93fb8f9db34ebb63eaef7ad783f56fc67d21871477d3952dbfa84b3527cea34a"
+    },
+    {
+      "path": "generation_config.json",
+      "bytes": 139,
+      "sha256": "215d4506cf90f3590fe821bee47dc92e99caa06f2c1b36076693b5209980009c"
+    },
+    {
+      "path": "merges.txt",
+      "bytes": 1671839,
+      "sha256": "599bab54075088774b1733fde865d5bd747cbcc7a547c5bc12610e874e26f5e3"
+    },
+    {
+      "path": "model.safetensors",
+      "bytes": 988097824,
+      "sha256": "6ff7b2528689b448ad8d06e0451ffae4f70f803bd32dc0ab8552f4a70c230e31"
+    },
+    {
+      "path": "provenance/model_lock.json",
+      "bytes": 3960,
+      "sha256": "8d844c172636648ed1e2503c61983c73fe03342c14bfe6f7de07e3714da1c547"
+    },
+    {
+      "path": "tokenizer.json",
+      "bytes": 7031645,
+      "sha256": "c0382117ea329cdf097041132f6d735924b697924d6f6fc3945713e96ce87539"
+    },
+    {
+      "path": "tokenizer_config.json",
+      "bytes": 7228,
+      "sha256": "283ec39d07ebfd1904aa2dee3c7d73524ed96c489fdb3915629900d3ab3ac319"
+    },
+    {
+      "path": "vocab.json",
+      "bytes": 2776833,
+      "sha256": "ca10d7e9fb3ed18575dd1e277a2579c16d108e32f27439684afa0e10b1440910"
+    }
+  ]
+}

tokenizer_config.json

@@ -0,0 +1,207 @@
+{
+  "add_prefix_space": false,
+  "added_tokens_decoder": {
+    "151643": {
+      "content": "<|endoftext|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151644": {
+      "content": "<|im_start|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151645": {
+      "content": "<|im_end|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151646": {
+      "content": "<|object_ref_start|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151647": {
+      "content": "<|object_ref_end|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151648": {
+      "content": "<|box_start|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151649": {
+      "content": "<|box_end|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151650": {
+      "content": "<|quad_start|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151651": {
+      "content": "<|quad_end|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151652": {
+      "content": "<|vision_start|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151653": {
+      "content": "<|vision_end|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151654": {
+      "content": "<|vision_pad|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151655": {
+      "content": "<|image_pad|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151656": {
+      "content": "<|video_pad|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151657": {
+      "content": "<tool_call>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151658": {
+      "content": "</tool_call>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151659": {
+      "content": "<|fim_prefix|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151660": {
+      "content": "<|fim_middle|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151661": {
+      "content": "<|fim_suffix|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151662": {
+      "content": "<|fim_pad|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151663": {
+      "content": "<|repo_name|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151664": {
+      "content": "<|file_sep|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    }
+  },
+  "additional_special_tokens": [
+    "<|im_start|>",
+    "<|im_end|>",
+    "<|object_ref_start|>",
+    "<|object_ref_end|>",
+    "<|box_start|>",
+    "<|box_end|>",
+    "<|quad_start|>",
+    "<|quad_end|>",
+    "<|vision_start|>",
+    "<|vision_end|>",
+    "<|vision_pad|>",
+    "<|image_pad|>",
+    "<|video_pad|>"
+  ],
+  "bos_token": null,
+  "chat_template": "{%- if tools %}\n    {{- '<|im_start|>system\\n' }}\n    {%- if messages[0]['role'] == 'system' %}\n        {{- messages[0]['content'] }}\n    {%- else %}\n        {{- 'You are a helpful assistant.' }}\n    {%- endif %}\n    {{- \"\\n\\n# Tools\\n\\nYou may call one or more functions to assist with the user query.\\n\\nYou are provided with function signatures within <tools></tools> XML tags:\\n<tools>\" }}\n    {%- for tool in tools %}\n        {{- \"\\n\" }}\n        {{- tool | tojson }}\n    {%- endfor %}\n    {{- \"\\n</tools>\\n\\nFor each function call, return a json object with function name and arguments within <tool_call></tool_call> XML tags:\\n<tool_call>\\n{\\\"name\\\": <function-name>, \\\"arguments\\\": <args-json-object>}\\n</tool_call><|im_end|>\\n\" }}\n{%- else %}\n    {%- if messages[0]['role'] == 'system' %}\n        {{- '<|im_start|>system\\n' + messages[0]['content'] + '<|im_end|>\\n' }}\n    {%- else %}\n        {{- '<|im_start|>system\\nYou are a helpful assistant.<|im_end|>\\n' }}\n    {%- endif %}\n{%- endif %}\n{%- for message in messages %}\n    {%- if (message.role == \"user\") or (message.role == \"system\" and not loop.first) or (message.role == \"assistant\" and not message.tool_calls) %}\n        {{- '<|im_start|>' + message.role + '\\n' + message.content + '<|im_end|>' + '\\n' }}\n    {%- elif message.role == \"assistant\" %}\n        {{- '<|im_start|>' + message.role }}\n        {%- if message.content %}\n            {{- '\\n' + message.content }}\n        {%- endif %}\n        {%- for tool_call in message.tool_calls %}\n            {%- if tool_call.function is defined %}\n                {%- set tool_call = tool_call.function %}\n            {%- endif %}\n            {{- '\\n<tool_call>\\n{\"name\": \"' }}\n            {{- tool_call.name }}\n            {{- '\", \"arguments\": ' }}\n            {{- tool_call.arguments | tojson }}\n            {{- '}\\n</tool_call>' }}\n        {%- endfor %}\n        {{- '<|im_end|>\\n' }}\n    {%- elif message.role == \"tool\" %}\n        {%- if (loop.index0 == 0) or (messages[loop.index0 - 1].role != \"tool\") %}\n            {{- '<|im_start|>user' }}\n        {%- endif %}\n        {{- '\\n<tool_response>\\n' }}\n        {{- message.content }}\n        {{- '\\n</tool_response>' }}\n        {%- if loop.last or (messages[loop.index0 + 1].role != \"tool\") %}\n            {{- '<|im_end|>\\n' }}\n        {%- endif %}\n    {%- endif %}\n{%- endfor %}\n{%- if add_generation_prompt %}\n    {{- '<|im_start|>assistant\\n' }}\n{%- endif %}\n",
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<|endoftext|>",
+  "errors": "replace",
+  "model_max_length": 32768,
+  "pad_token": "<|endoftext|>",
+  "split_special_tokens": false,
+  "tokenizer_class": "Qwen2Tokenizer",
+  "unk_token": null,
+  "add_bos_token": false
+}