Add coding-50-nonuniform

.gitattributes

@@ -36,3 +36,4 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
 coding-25-uniform/tokenizer.json filter=lfs diff=lfs merge=lfs -text
 coding-25-nonuniform/tokenizer.json filter=lfs diff=lfs merge=lfs -text
 coding-50-uniform/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+coding-50-nonuniform/tokenizer.json filter=lfs diff=lfs merge=lfs -text

coding-50-nonuniform/LOAD_VLLM.md

@@ -0,0 +1,43 @@
+# Loading this variant with vLLM
+
+The checkpoint has heterogeneous per-layer expert counts. vLLM's stock
+`Qwen3NextSparseMoeBlock` builds every layer with `config.num_experts`; our
+bundled `vllm_pruned_patch.py` overrides each layer to use its own count from
+`config.per_layer_num_experts`.
+
+## One-liner
+
+```bash
+PYTHONPATH=$(pwd):${PYTHONPATH:-} python -c "
+from vllm import LLM, SamplingParams
+llm = LLM(model='.', tensor_parallel_size=4, dtype='bfloat16',
+          gpu_memory_utilization=0.85, trust_remote_code=True,
+          enforce_eager=True)
+print(llm.generate(['def fib(n):'], SamplingParams(max_tokens=128))[0].outputs[0].text)
+"
+```
+
+## Why PYTHONPATH?
+
+vLLM spawns worker subprocesses via `multiprocessing.spawn` (safe with CUDA).
+Those workers re-import `vllm` fresh — any monkey-patch you applied in the
+parent process is gone. Python's `sitecustomize.py` mechanism runs
+automatically in **every** interpreter that has the relevant directory on
+`sys.path`. Putting this variant folder on `PYTHONPATH` is enough.
+
+## Tensor parallelism notes
+
+- **TP (tensor parallel)** works fine with heterogeneous counts — TP shards
+  hidden dimensions inside each expert.
+- **EP (expert parallel)** assumes experts shard evenly across ranks, which
+  is broken for heterogeneous counts. Keep `--enable-eplb` off.
+
+## With lm-eval-harness
+
+```bash
+PYTHONPATH=$(pwd):${PYTHONPATH:-} \
+  lm_eval --model vllm \
+    --model_args "pretrained=.,tensor_parallel_size=4,dtype=bfloat16,gpu_memory_utilization=0.85,max_model_len=4096,trust_remote_code=True,enforce_eager=True" \
+    --tasks humaneval,mbpp \
+    --batch_size auto
+```

coding-50-nonuniform/added_tokens.json

@@ -0,0 +1,28 @@
+{
+  "</think>": 151668,
+  "</tool_call>": 151658,
+  "</tool_response>": 151666,
+  "<think>": 151667,
+  "<tool_call>": 151657,
+  "<tool_response>": 151665,
+  "<|box_end|>": 151649,
+  "<|box_start|>": 151648,
+  "<|endoftext|>": 151643,
+  "<|file_sep|>": 151664,
+  "<|fim_middle|>": 151660,
+  "<|fim_pad|>": 151662,
+  "<|fim_prefix|>": 151659,
+  "<|fim_suffix|>": 151661,
+  "<|im_end|>": 151645,
+  "<|im_start|>": 151644,
+  "<|image_pad|>": 151655,
+  "<|object_ref_end|>": 151647,
+  "<|object_ref_start|>": 151646,
+  "<|quad_end|>": 151651,
+  "<|quad_start|>": 151650,
+  "<|repo_name|>": 151663,
+  "<|video_pad|>": 151656,
+  "<|vision_end|>": 151653,
+  "<|vision_pad|>": 151654,
+  "<|vision_start|>": 151652
+}

coding-50-nonuniform/chat_template.jinja

@@ -0,0 +1,117 @@
+{% macro render_extra_keys(json_dict, handled_keys) %}
+    {%- if json_dict is mapping %}
+        {%- for json_key in json_dict if json_key not in handled_keys %}
+            {%- if json_dict[json_key] is string %}
+                {{-'\n<' ~ json_key ~ '>' ~ (json_dict[json_key] | string) ~ '</' ~ json_key ~ '>' }}
+            {%- else %}
+                {{- '\n<' ~ json_key ~ '>' ~ (json_dict[json_key] | tojson | safe) ~ '</' ~ json_key ~ '>' }}
+            {%- endif %}
+        {%- endfor %}
+    {%- endif %}
+{%- endmacro %}
+
+{%- if messages[0]["role"] == "system" %}
+    {%- set system_message = messages[0]["content"] %}
+    {%- set loop_messages = messages[1:] %}
+{%- else %}
+    {%- set loop_messages = messages %}
+{%- endif %}
+
+{%- if not tools is defined %}
+    {%- set tools = [] %}
+{%- endif %}
+
+{%- if system_message is defined %}
+    {{- "<|im_start|>system\n" + system_message }}
+{%- else %}
+    {%- if tools is iterable and tools | length > 0 %}
+        {{- "<|im_start|>system\nYou are Qwen, a helpful AI assistant that can interact with a computer to solve tasks." }}
+    {%- endif %}
+{%- endif %}
+{%- if tools is iterable and tools | length > 0 %}
+    {{- "\n\n# Tools\n\nYou have access to the following functions:\n\n" }}
+    {{- "<tools>" }}
+    {%- for tool in tools %}
+        {%- if tool.function is defined %}
+            {%- set tool = tool.function %}
+        {%- endif %}
+        {{- "\n<function>\n<name>" ~ tool.name ~ "</name>" }}
+        {%- if tool.description is defined %}
+            {{- '\n<description>' ~ (tool.description | trim) ~ '</description>' }}
+        {%- endif %}
+        {{- '\n<parameters>' }}
+        {%- if tool.parameters is defined and tool.parameters is mapping and tool.parameters.properties is defined and tool.parameters.properties is mapping %}
+            {%- for param_name, param_fields in tool.parameters.properties|items %}
+                {{- '\n<parameter>' }}
+                {{- '\n<name>' ~ param_name ~ '</name>' }}
+                {%- if param_fields.type is defined %}
+                    {{- '\n<type>' ~ (param_fields.type | string) ~ '</type>' }}
+                {%- endif %}
+                {%- if param_fields.description is defined %}
+                    {{- '\n<description>' ~ (param_fields.description | trim) ~ '</description>' }}
+                {%- endif %}
+                {%- set handled_keys = ['name', 'type', 'description'] %}
+                {{- render_extra_keys(param_fields, handled_keys) }}
+                {{- '\n</parameter>' }}
+            {%- endfor %}
+        {%- endif %}
+        {%- set handled_keys = ['type', 'properties'] %}
+        {{- render_extra_keys(tool.parameters, handled_keys) }}
+        {{- '\n</parameters>' }}
+        {%- set handled_keys = ['type', 'name', 'description', 'parameters'] %}
+        {{- render_extra_keys(tool, handled_keys) }}
+        {{- '\n</function>' }}
+    {%- endfor %}
+    {{- "\n</tools>" }}
+    {{- '\n\nIf you choose to call a function ONLY reply in the following format with NO suffix:\n\n<tool_call>\n<function=example_function_name>\n<parameter=example_parameter_1>\nvalue_1\n</parameter>\n<parameter=example_parameter_2>\nThis is the value for the second parameter\nthat can span\nmultiple lines\n</parameter>\n</function>\n</tool_call>\n\n<IMPORTANT>\nReminder:\n- Function calls MUST follow the specified format: an inner <function=...></function> block must be nested within <tool_call></tool_call> XML tags\n- Required parameters MUST be specified\n- You may provide optional reasoning for your function call in natural language BEFORE the function call, but NOT after\n- If there is no function call available, answer the question like normal with your current knowledge and do not tell the user about function calls\n</IMPORTANT>' }}
+{%- endif %}
+{%- if system_message is defined %}
+    {{- '<|im_end|>\n' }}
+{%- else %}
+    {%- if tools is iterable and tools | length > 0 %}
+        {{- '<|im_end|>\n' }}
+    {%- endif %}
+{%- endif %}
+{%- for message in loop_messages %}
+    {%- if message.role == "assistant" and message.tool_calls is defined and message.tool_calls is iterable and message.tool_calls | length > 0 %}
+        {{- '<|im_start|>' + message.role }}
+        {%- if message.content is defined and message.content is string and message.content | trim | length > 0 %}
+            {{- '\n' + message.content | trim + '\n' }}
+        {%- endif %}
+        {%- for tool_call in message.tool_calls %}
+            {%- if tool_call.function is defined %}
+                {%- set tool_call = tool_call.function %}
+            {%- endif %}
+            {{- '\n<tool_call>\n<function=' + tool_call.name + '>\n' }}
+            {%- if tool_call.arguments is defined %}
+                {%- for args_name, args_value in tool_call.arguments|items %}
+                    {{- '<parameter=' + args_name + '>\n' }}
+                    {%- set args_value = args_value if args_value is string else args_value | tojson | safe %}
+                    {{- args_value }}
+                    {{- '\n</parameter>\n' }}
+                {%- endfor %}
+            {%- endif %}
+            {{- '</function>\n</tool_call>' }}
+        {%- endfor %}
+        {{- '<|im_end|>\n' }}
+    {%- elif message.role == "user" or message.role == "system" or message.role == "assistant" %}
+        {{- '<|im_start|>' + message.role + '\n' + message.content + '<|im_end|>' + '\n' }}
+    {%- elif message.role == "tool" %}
+        {%- if loop.previtem and loop.previtem.role != "tool" %}
+            {{- '<|im_start|>user' }}
+        {%- endif %}
+        {{- '\n<tool_response>\n' }}
+        {{- message.content }}
+        {{- '\n</tool_response>' }}
+        {%- if not loop.last and loop.nextitem.role != "tool" %}
+            {{- '<|im_end|>\n' }}
+        {%- elif loop.last %}
+            {{- '<|im_end|>\n' }}
+        {%- endif %}
+    {%- else %}
+        {{- '<|im_start|>' + message.role + '\n' + message.content + '<|im_end|>\n' }}
+    {%- endif %}
+{%- endfor %}
+{%- if add_generation_prompt %}
+    {{- '<|im_start|>assistant\n' }}
+{%- endif %}

coding-50-nonuniform/config.json

@@ -0,0 +1,145 @@
+{
+  "architectures": [
+    "Qwen3NextForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0,
+  "bos_token_id": 151643,
+  "decoder_sparse_step": 1,
+  "dtype": "bfloat16",
+  "eos_token_id": 151645,
+  "full_attention_interval": 4,
+  "head_dim": 256,
+  "hidden_act": "silu",
+  "hidden_size": 2048,
+  "initializer_range": 0.02,
+  "intermediate_size": 5120,
+  "layer_types": [
+    "linear_attention",
+    "linear_attention",
+    "linear_attention",
+    "full_attention",
+    "linear_attention",
+    "linear_attention",
+    "linear_attention",
+    "full_attention",
+    "linear_attention",
+    "linear_attention",
+    "linear_attention",
+    "full_attention",
+    "linear_attention",
+    "linear_attention",
+    "linear_attention",
+    "full_attention",
+    "linear_attention",
+    "linear_attention",
+    "linear_attention",
+    "full_attention",
+    "linear_attention",
+    "linear_attention",
+    "linear_attention",
+    "full_attention",
+    "linear_attention",
+    "linear_attention",
+    "linear_attention",
+    "full_attention",
+    "linear_attention",
+    "linear_attention",
+    "linear_attention",
+    "full_attention",
+    "linear_attention",
+    "linear_attention",
+    "linear_attention",
+    "full_attention",
+    "linear_attention",
+    "linear_attention",
+    "linear_attention",
+    "full_attention",
+    "linear_attention",
+    "linear_attention",
+    "linear_attention",
+    "full_attention",
+    "linear_attention",
+    "linear_attention",
+    "linear_attention",
+    "full_attention"
+  ],
+  "linear_conv_kernel_dim": 4,
+  "linear_key_head_dim": 128,
+  "linear_num_key_heads": 16,
+  "linear_num_value_heads": 32,
+  "linear_value_head_dim": 128,
+  "max_position_embeddings": 262144,
+  "mlp_only_layers": [],
+  "model_type": "qwen3_next",
+  "moe_intermediate_size": 512,
+  "norm_topk_prob": true,
+  "num_attention_heads": 16,
+  "num_experts": 512,
+  "num_experts_per_tok": 10,
+  "num_hidden_layers": 48,
+  "num_key_value_heads": 2,
+  "original_num_experts": 512,
+  "output_router_logits": false,
+  "partial_rotary_factor": 0.25,
+  "per_layer_num_experts": [
+    359,
+    279,
+    365,
+    194,
+    214,
+    270,
+    287,
+    272,
+    245,
+    212,
+    218,
+    231,
+    188,
+    213,
+    236,
+    272,
+    208,
+    281,
+    278,
+    265,
+    236,
+    227,
+    223,
+    250,
+    205,
+    230,
+    253,
+    288,
+    217,
+    289,
+    289,
+    281,
+    226,
+    213,
+    215,
+    238,
+    217,
+    226,
+    267,
+    303,
+    304,
+    297,
+    290,
+    285,
+    296,
+    269,
+    267,
+    300
+  ],
+  "rms_norm_eps": 1e-06,
+  "rope_scaling": null,
+  "rope_theta": 5000000,
+  "router_aux_loss_coef": 0.001,
+  "shared_expert_intermediate_size": 512,
+  "tie_word_embeddings": false,
+  "transformers_version": "4.57.6",
+  "use_cache": true,
+  "use_sliding_window": false,
+  "vocab_size": 151936
+}

coding-50-nonuniform/generation_config.json

@@ -0,0 +1,12 @@
+{
+  "bos_token_id": 151643,
+  "do_sample": true,
+  "eos_token_id": [
+    151645,
+    151643
+  ],
+  "pad_token_id": 151643,
+  "top_k": 40,
+  "top_p": 0.95,
+  "transformers_version": "4.57.6"
+}

coding-50-nonuniform/model-00001-of-00017.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:460b6bb9546b46956866c36342968e5d34fce98c1e1c1f23b17785659d30fa27
+size 4998161384

coding-50-nonuniform/model-00002-of-00017.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4eb89e722e2d58608bb8fbf76fbe6f4811056b2a71d2e27c764b7446ad33db64
+size 4999173848

coding-50-nonuniform/model-00003-of-00017.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:027701bf129d467c411240cc495c9d67914047d27c297651659beab0ec47e091
+size 4998348808

coding-50-nonuniform/model-00004-of-00017.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:98fa3d2f7d949f647f59cfa120e9b3e6eb7ef434d9bc08527bed34c56a062c1f
+size 4998348800

coding-50-nonuniform/model-00005-of-00017.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:657581489d31d5471ce9e27622bb6eed6edbac919e598e45523179968dc77390
+size 4999334400

coding-50-nonuniform/model-00006-of-00017.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:585cec3d65a5390bb0f5d4f277644472a68549c07684b6f084e24c1aab81ab26
+size 4999516104

coding-50-nonuniform/model-00007-of-00017.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2c44e668e78786792f6495b2874b1498e0cef56d9bbad21e1a38e9a60c6ccd88
+size 4999589880

coding-50-nonuniform/model-00008-of-00017.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3341924b1799e44b828059935d99c998f5ac9d6a30b7ac21da7a1ffb8f752656
+size 4999266296

coding-50-nonuniform/model-00009-of-00017.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:963623c94fac0e6bd5f40bf5772cd24cae14eb6a596118b4e606c32e2acb1ee3
+size 4998637040

coding-50-nonuniform/model-00010-of-00017.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f1d5d95997b1018037949d19121db2ce868280612a552b5cab7d3a24e8c8cbde
+size 4998460512

coding-50-nonuniform/model-00011-of-00017.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:dbd88c027716ab256ee1033c0c588a8994a33f82dbda4a99fe2e73a1775b316b
+size 4999659544

coding-50-nonuniform/model-00012-of-00017.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5e77d4e9768c6c43e6bc97b645c5acd93c8f911ba2ea42c4cfbb9a8ff1403609
+size 4998256040

coding-50-nonuniform/model-00013-of-00017.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:41689f095b5fb5c30f28e1470f2c8269ad8a0828be3282a886b9056df23f8936
+size 4999659312

coding-50-nonuniform/model-00014-of-00017.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:faf73ab3a2d59aac12fb077c14975538520d8f6bf6327f170ee06c535be8ea06
+size 4998849856

coding-50-nonuniform/model-00015-of-00017.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ad9af69804d6a3228cb168bbfc460c112c3d9f0dbe67efdc641f0e7f3300c2c3
+size 4999966712

coding-50-nonuniform/model-00016-of-00017.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6a6823a90add18b91497c570ecefeade53457cd9fae3ee799263788db1f9cec1
+size 4999823408

coding-50-nonuniform/model-00017-of-00017.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3f40332c93907947014981154488dde2841e826ccfc1637e99b70256c3f13c9e
+size 2008646720

coding-50-nonuniform/pruned_metadata.json

@@ -0,0 +1,60 @@
+{
+  "base_model": "Qwen/Qwen3-Coder-Next",
+  "prune_mask": "/nfs/scistore19/alistgrp/mhelcig/repos/rco/llm/results/coder_next/mask_perexpert_reap_50pct.pt",
+  "original_num_experts": 512,
+  "num_hidden_layers": 48,
+  "num_experts_per_tok": 10,
+  "per_layer_num_experts": [
+    359,
+    279,
+    365,
+    194,
+    214,
+    270,
+    287,
+    272,
+    245,
+    212,
+    218,
+    231,
+    188,
+    213,
+    236,
+    272,
+    208,
+    281,
+    278,
+    265,
+    236,
+    227,
+    223,
+    250,
+    205,
+    230,
+    253,
+    288,
+    217,
+    289,
+    289,
+    281,
+    226,
+    213,
+    215,
+    238,
+    217,
+    226,
+    267,
+    303,
+    304,
+    297,
+    290,
+    285,
+    296,
+    269,
+    267,
+    300
+  ],
+  "total_pruned": 12288,
+  "sparsity": 0.5,
+  "allocation": "nonuniform"
+}

coding-50-nonuniform/sitecustomize.py

@@ -0,0 +1,12 @@
+"""Applied automatically in every Python process on PYTHONPATH.
+vLLM spawns worker subprocesses that re-import modules; without this file
+the monkey-patch would be missing in workers and weight loading would fail
+for the heterogeneous per-layer expert counts.
+"""
+import os
+if os.environ.get('VLLM_PRUNED_PATCH_DISABLE') != '1':
+    try:
+        import vllm_pruned_patch
+        vllm_pruned_patch.apply()
+    except Exception:
+        pass

coding-50-nonuniform/special_tokens_map.json

@@ -0,0 +1,31 @@
+{
+  "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|>"
+  ],
+  "eos_token": {
+    "content": "<|im_end|>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "pad_token": {
+    "content": "<|endoftext|>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  }
+}

coding-50-nonuniform/tokenizer.json

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:aeb13307a71acd8fe81861d94ad54ab689df773318809eed3cbe794b4492dae4
+size 11422654

coding-50-nonuniform/tokenizer_config.json

@@ -0,0 +1,239 @@
+{
+  "add_bos_token": false,
+  "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
+    },
+    "151665": {
+      "content": "<tool_response>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151666": {
+      "content": "</tool_response>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151667": {
+      "content": "<think>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151668": {
+      "content": "</think>",
+      "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,
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<|im_end|>",
+  "errors": "replace",
+  "extra_special_tokens": {},
+  "model_max_length": 1048576,
+  "pad_token": "<|endoftext|>",
+  "split_special_tokens": false,
+  "tokenizer_class": "Qwen2Tokenizer",
+  "unk_token": null
+}

coding-50-nonuniform/vllm_pruned_patch.py

@@ -0,0 +1,88 @@
+"""Monkey-patch vLLM's Qwen3-Next MoE block to support heterogeneous per-layer expert counts.
+
+Use:
+    import vllm_pruned_patch
+    vllm_pruned_patch.apply()
+    # ...then any vLLM import / engine / lcb_runner call
+
+The patch reads `config.per_layer_num_experts` (a list of length num_hidden_layers)
+written by `prune_and_save.py`. If that field is missing the patch is a no-op,
+so it's safe to apply unconditionally.
+
+Tested against vLLM 0.16.0 / Qwen3NextSparseMoeBlock.
+EP (expert parallelism) is NOT supported with heterogeneous counts —
+keep --enable-eplb off (the default) and stick to TP-only.
+"""
+
+from __future__ import annotations
+
+import logging
+
+logger = logging.getLogger(__name__)
+
+_PATCHED = False
+
+
+def apply():
+    """Idempotently install the patch. Safe to call multiple times."""
+    global _PATCHED
+    if _PATCHED:
+        return
+
+    from vllm.model_executor.models import qwen3_next as qm
+    from vllm.model_executor.models.utils import extract_layer_index
+
+    OriginalBlock = qm.Qwen3NextSparseMoeBlock
+    original_init = OriginalBlock.__init__
+
+    def patched_init(self, vllm_config, prefix: str = ""):
+        cfg = vllm_config.model_config.hf_config
+        per_layer = getattr(cfg, "per_layer_num_experts", None)
+        if per_layer is None:
+            # No pruning metadata — fall through to original behavior.
+            return original_init(self, vllm_config, prefix=prefix)
+
+        layer_idx = extract_layer_index(prefix)
+        n_local = int(per_layer[layer_idx])
+        original_n = cfg.num_experts
+
+        # Temporarily override num_experts on the (shared) hf_config so the
+        # original __init__ constructs gate + FusedMoE with the right size.
+        # Restore immediately after to avoid bleeding into other layers.
+        try:
+            cfg.num_experts = n_local
+            original_init(self, vllm_config, prefix=prefix)
+        finally:
+            cfg.num_experts = original_n
+
+    qm.Qwen3NextSparseMoeBlock.__init__ = patched_init
+
+    # Patch get_expert_mapping to use max kept count (smaller mapping table).
+    OriginalLM = qm.Qwen3NextForCausalLM
+    original_get_expert_mapping = OriginalLM.get_expert_mapping
+
+    def patched_get_expert_mapping(self):
+        per_layer = getattr(self.config, "per_layer_num_experts", None)
+        if per_layer is None:
+            return original_get_expert_mapping(self)
+        from vllm.model_executor.layers.fused_moe.shared_fused_moe import (
+            SharedFusedMoE,
+        )
+        # Saved checkpoint always has expert ids 0..max-1 across layers
+        # (each layer is re-indexed densely in prune_and_save.py).
+        max_n = max(int(x) for x in per_layer)
+        return SharedFusedMoE.make_expert_params_mapping(
+            self,
+            ckpt_gate_proj_name="gate_proj",
+            ckpt_down_proj_name="down_proj",
+            ckpt_up_proj_name="up_proj",
+            num_experts=max_n,
+            num_redundant_experts=self.num_redundant_experts,
+        )
+
+    qm.Qwen3NextForCausalLM.get_expert_mapping = patched_get_expert_mapping
+
+    _PATCHED = True
+    logger.info(
+        "Applied Qwen3-Next pruned-model patch (heterogeneous per-layer expert counts)."
+    )