Piko-9b
Copyright 2026 Dexy

This product includes software and model weights developed by third parties.
Piko-9b is not trained from scratch: its published weights are a tensor-level
composition of two upstream checkpoints. Both upstream licences apply to the
corresponding portions of the weights and must be preserved in redistribution.

================================================================================
1. Language backbone — 9,197,093,888 parameters (427 tensors)
================================================================================

Derived, through a chain of QLoRA fine-tunes, from:

    deepreinforce-ai/Ornith-1.0-9B
    https://huggingface.co/deepreinforce-ai/Ornith-1.0-9B
    Licensed under the MIT License.

MIT License

Permission is hereby granted, free of charge, to any person obtaining a copy
of this software and associated documentation files (the "Software"), to deal
in the Software without restriction, including without limitation the rights
to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
copies of the Software, and to permit persons to whom the Software is
furnished to do so, subject to the following conditions:

The above copyright notice and this permission notice shall be included in all
copies or substantial portions of the Software.

THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
SOFTWARE.

================================================================================
2. Vision tower and merger — 456,010,480 parameters (348 tensors)
================================================================================

Copied verbatim, without modification or further training, from:

    Qwen/Qwen3.5-9B
    https://huggingface.co/Qwen/Qwen3.5-9B
    Copyright 2026 Alibaba Cloud / Qwen Team
    Licensed under the Apache License, Version 2.0.

================================================================================
3. Tokenizer, vocabulary, chat template and processor configuration
================================================================================

Derived from the Qwen3.5 tokenizer and processor configuration
(Qwen/Qwen3.5-9B), Apache License, Version 2.0.

================================================================================
Licence compatibility
================================================================================

The MIT License is compatible with redistribution of the combined work under
the Apache License, Version 2.0, provided the MIT copyright notice and
permission notice above are retained. That is the purpose of this file.

You may obtain a copy of the Apache 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.

================================================================================
Training data
================================================================================

The QLoRA fine-tuning stages that produced the language backbone used datasets
whose individual licences are not fully documented in this repository. Dataset
manifests record category names and row counts, but several upstream sources
are identified only by local filename. This is recorded as an unresolved
provenance gap in reports/repository_audit.md rather than asserted to be clear.