Upload folder using huggingface_hub

.gitattributes

@@ -33,3 +33,4 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
 *.zip filter=lfs diff=lfs merge=lfs -text
 *.zst filter=lfs diff=lfs merge=lfs -text
 *tfevents* filter=lfs diff=lfs merge=lfs -text
+assets/ubtree_pipeline.png filter=lfs diff=lfs merge=lfs -text

README.md

@@ -1,3 +1,110 @@
 ---
+library_name: sglang
+pipeline_tag: text-generation
 license: mit
+base_model:
+  - inclusionAI/Ling-3.0-tiny
+tags:
+  - speculative-decoding
+  - tree-speculative-decoding
+  - ubtree
+  - dflash
+  - sglang
+  - ling3
+  - arxiv:2609.39972
+language:
+  - zh
+  - en
 ---
+
+# Ling3-Tiny-UBTree-B16
+
+[Paper](https://arxiv.org/abs/2609.39972) · [Target model](https://huggingface.co/inclusionAI/Ling-3.0-tiny)
+
+**Ling3-Tiny-UBTree-B16** is a UBTree speculative-decoding draft model for
+[inclusionAI/Ling-3.0-tiny](https://huggingface.co/inclusionAI/Ling-3.0-tiny). It combines a
+five-layer DFlash backbone with a learned bigram selector to propose multiple
+candidate continuations for target-model verification.
+
+> **SGLang integration — stay tuned!** Our UBTree adaptation for Ling3 has not
+> yet been released to the open-source SGLang community. Public installation
+> instructions and serving recipes will be added here when it becomes available.
+
+This checkpoint is used alongside the matching target model; it cannot serve
+as a standalone text-generation model.
+
+## How UBTree works
+
+![UBTree architecture and tree-native training pipeline](assets/ubtree_pipeline.png)
+
+*UBTree pipeline*
+
+UBTree pairs a **unigram proposer** with a **bigram selector**:
+
+1. A DFlash backbone proposes candidate tokens for future positions in parallel.
+2. The selector scores transitions between adjacent candidates in parallel.
+3. Tree construction retains promising branches within a verification budget.
+4. The target verifies the tree and commits the accepted continuation.
+
+The paper also introduces tree-native training with renormalized KL supervision
+on high-temperature target traces, encouraging plausible alternative branches.
+See the [paper](https://arxiv.org/abs/2609.39972) for the method and experiments.
+
+## Model specifications
+
+| Item | Value |
+|---|---|
+| Target model | [inclusionAI/Ling-3.0-tiny](https://huggingface.co/inclusionAI/Ling-3.0-tiny) |
+| Draft architecture | `Ling3UBTreeModel` |
+| Draft backbone | 5 full-attention Transformer layers |
+| Parameters stored in this checkpoint | 183,127,326 (183.1M) |
+| Weight dtype | BF16 |
+| Hidden size | 1,536 |
+| Attention heads / KV heads | 16 / 16 |
+| Target auxiliary feature layers (zero-based) | 4, 8, 12, 16, 22 |
+| Draft block size | 16 (anchor-inclusive runtime convention) |
+| Candidate top-k | 64 per draft position |
+| Default verification tree budget | 64 non-root nodes + 1 anchor |
+| Training | DFlash warm start, then joint backbone + UBTree selector training |
+
+The parameter count covers tensors distributed in this repository. The draft
+shares token embeddings and the language-model head with the target; these
+shared weights are not included here. B16 denotes the draft block size, not
+the verification tree's node budget.
+
+## Serving with SGLang
+
+UBTree support for Ling3 in SGLang is coming soon. We will share installation
+instructions and serving examples once the integration is publicly available.
+**Stay tuned!**
+
+## Evaluation notes
+
+The [UBTree paper](https://arxiv.org/abs/2609.39972) reports results under its
+specified models and experimental configurations. Those results should not
+be treated as measurements of every released checkpoint or serving setup.
+
+Runtime version, hardware, tree budget, concurrency, and decoding settings
+all affect acceptance length and speed. Checkpoint-specific reproducible
+benchmark recipes will accompany the public runtime integration. Numerical
+or token-level equivalence and task quality require validation for the chosen
+serving configuration.
+
+## Citation
+
+If you use UBTree in your work, please cite:
+
+```bibtex
+@article{liang2026ubtree,
+  title   = {UBTree: Parallel Tree Drafting via Unigram and Bigram Models for Speculative Decoding},
+  author  = {Liang, Chumeng and Wang, Linxuan and Peng, Xinyu and Liu, Huabin and Chen, Yuxin and Liu, Ge and Lin, Guang and Song, Qifan and Li, Jianguo},
+  journal = {arXiv preprint arXiv:2609.39972},
+  year    = {2026},
+  url     = {https://arxiv.org/abs/2609.39972}
+}
+```
+
+## License
+
+The model weights are released under the **MIT License**. The reproduced paper
+figure is separately licensed under **CC BY 4.0**, as credited above.

config.json

@@ -0,0 +1,147 @@
+{
+  "architectures": [
+    "Ling3UBTreeModel"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "block_size": 16,
+  "bos_token_id": null,
+  "dflash_config": {
+    "block_size": 16,
+    "contract": "flashflow_anchor_inclusive_v1",
+    "mask_token_id": 156892,
+    "num_target_layers": 24,
+    "target_layer_ids": [
+      4,
+      8,
+      12,
+      16,
+      22
+    ]
+  },
+  "dflash_contract": "flashflow_anchor_inclusive_v1",
+  "dtype": "bfloat16",
+  "embedding_dropout": 0.0,
+  "enable_confidence_head": false,
+  "eos_token_id": 156895,
+  "expert_swiglu_limit_list": null,
+  "first_k_dense_replace": 1,
+  "gated_attention_proj_granularity_type": "head_wise",
+  "group_norm_size": 1,
+  "head_dim": 128,
+  "hidden_act": "silu",
+  "hidden_size": 1536,
+  "initializer_range": 0.02,
+  "intermediate_size": 4608,
+  "kda_lower_bound": -5,
+  "kda_safe_gate": true,
+  "kv_lora_rank": 512,
+  "layer_group_size": 4,
+  "layer_types": [
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention"
+  ],
+  "linear_silu": true,
+  "markov_rank": 0,
+  "mask_token_id": 156892,
+  "max_position_embeddings": 131072,
+  "max_window_layers": 20,
+  "model_type": "qwen3",
+  "moe_intermediate_size": 512,
+  "moe_router_enable_expert_bias": true,
+  "moe_shared_expert_intermediate_size": 512,
+  "mtp_loss_scaling_factor": 0,
+  "mtp_use_kda": false,
+  "n_group": 8,
+  "no_kda_lora": true,
+  "norm_topk_prob": true,
+  "num_anchors": 512,
+  "num_attention_heads": 16,
+  "num_experts": 128,
+  "num_experts_per_tok": 8,
+  "num_hidden_layers": 5,
+  "num_key_value_heads": 16,
+  "num_kv_heads_for_linear_attn": 0,
+  "num_shared_experts": 1,
+  "num_target_layers": 24,
+  "output_dropout": 0.0,
+  "output_router_logits": false,
+  "pad_token_id": 156892,
+  "partial_rotary_factor": 1.0,
+  "q_lora_rank": 256,
+  "qk_head_dim": 192,
+  "qk_nope_head_dim": 128,
+  "qk_rope_head_dim": 64,
+  "rms_norm_eps": 1e-06,
+  "rope_interleave": true,
+  "rope_parameters": {
+    "partial_rotary_factor": 1.0,
+    "rope_theta": 6000000,
+    "rope_type": "default"
+  },
+  "rope_theta": 6000000.0,
+  "rotary_dim": 64,
+  "routed_scaling_factor": 2.5,
+  "router_dtype": "fp32",
+  "scale_router_input": false,
+  "score_function": "sigmoid",
+  "scoring_func": "sigmoid",
+  "self_conditioning": false,
+  "seq_aux": true,
+  "share_expert_swiglu_limit_list": null,
+  "short_conv_kernel_size": 4,
+  "sliding_window": null,
+  "target_layer_ids": [
+    4,
+    8,
+    12,
+    16,
+    22
+  ],
+  "tie_word_embeddings": false,
+  "topk_group": 4,
+  "topk_method": "noaux_tc",
+  "transformers_version": "5.12.1",
+  "ubtree_config": {
+    "adds_unary_logits": true,
+    "architecture": "cp",
+    "candidate_k": 64,
+    "candidate_source": "frozen_dflash_unary_topk_ids_only",
+    "depth_condition_count": 15,
+    "depth_conditioning": "unary_correction_two_scale",
+    "depth_scale_parameterization": "exp_log_scale",
+    "factorizer_hidden_width": 512,
+    "hidden_projection_depth_count": 15,
+    "hidden_projection_type": "shared",
+    "inference_hidden_projection": "materialized_per_depth_affine",
+    "kl_temperature": 1.0,
+    "objective": "kl",
+    "rank": 256,
+    "target_embedding_dim": 1536,
+    "token_factorizer": "target_embedding_mlp",
+    "train_backbone": true,
+    "trained": true,
+    "training_mode": "joint_backbone_selector",
+    "tree_budget": 64,
+    "tree_builder": "pruned",
+    "tree_depth_bonus": 0.0,
+    "tree_supertree_width": 12,
+    "type": "residual_bigram",
+    "version": 1
+  },
+  "up_proj_norm": false,
+  "use_bias": false,
+  "use_cache": true,
+  "use_kda_lora": false,
+  "use_mla_nope": false,
+  "use_nGPT": false,
+  "use_qk_norm": true,
+  "use_qkv_bias": false,
+  "use_sliding_window": false,
+  "v_head_dim": 128,
+  "value_norm": false,
+  "vocab_size": 157184
+}

model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c0820dfec7492fd6543dc170be8db2de076bc88524cf7d3dce1de37fe3173dca
+size 366261868