Add files using upload-large-folder tool

Qwen3-32B/ll_bsearch_kl0.005_eap0.99_sha_bw5.53_4-5-6-7-8bit_grouped_seed42/README.md

@@ -0,0 +1,18 @@
+# Quantized Model Checkpoint
+
+**Base model:** Qwen/Qwen3-32B
+
+**Average bitwidth:** 5.5305
+
+**Sensitivity method:** shapley
+
+**Constraints:**
+- max_kl: 0.005
+- min_eap: 0.99
+
+**Metrics:**
+- final_kl: 0.001976
+- final_eap: 0.990013
+- final_etl: 0.009987
+
+See `quantization_config.txt` for full configuration details.

Qwen3-32B/ll_bsearch_kl0.005_eap0.99_sha_bw5.53_4-5-6-7-8bit_grouped_seed42/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
+}

Qwen3-32B/ll_bsearch_kl0.005_eap0.99_sha_bw5.53_4-5-6-7-8bit_grouped_seed42/chat_template.jinja

@@ -0,0 +1,89 @@
+{%- if tools %}
+    {{- '<|im_start|>system\n' }}
+    {%- if messages[0].role == 'system' %}
+        {{- messages[0].content + '\n\n' }}
+    {%- endif %}
+    {{- "# 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>" }}
+    {%- for tool in tools %}
+        {{- "\n" }}
+        {{- tool | tojson }}
+    {%- endfor %}
+    {{- "\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" }}
+{%- else %}
+    {%- if messages[0].role == 'system' %}
+        {{- '<|im_start|>system\n' + messages[0].content + '<|im_end|>\n' }}
+    {%- endif %}
+{%- endif %}
+{%- set ns = namespace(multi_step_tool=true, last_query_index=messages|length - 1) %}
+{%- for message in messages[::-1] %}
+    {%- set index = (messages|length - 1) - loop.index0 %}
+    {%- if ns.multi_step_tool and message.role == "user" and message.content is string and not(message.content.startswith('<tool_response>') and message.content.endswith('</tool_response>')) %}
+        {%- set ns.multi_step_tool = false %}
+        {%- set ns.last_query_index = index %}
+    {%- endif %}
+{%- endfor %}
+{%- for message in messages %}
+    {%- if message.content is string %}
+        {%- set content = message.content %}
+    {%- else %}
+        {%- set content = '' %}
+    {%- endif %}
+    {%- if (message.role == "user") or (message.role == "system" and not loop.first) %}
+        {{- '<|im_start|>' + message.role + '\n' + content + '<|im_end|>' + '\n' }}
+    {%- elif message.role == "assistant" %}
+        {%- set reasoning_content = '' %}
+        {%- if message.reasoning_content is string %}
+            {%- set reasoning_content = message.reasoning_content %}
+        {%- else %}
+            {%- if '</think>' in content %}
+                {%- set reasoning_content = content.split('</think>')[0].rstrip('\n').split('<think>')[-1].lstrip('\n') %}
+                {%- set content = content.split('</think>')[-1].lstrip('\n') %}
+            {%- endif %}
+        {%- endif %}
+        {%- if loop.index0 > ns.last_query_index %}
+            {%- if loop.last or (not loop.last and reasoning_content) %}
+                {{- '<|im_start|>' + message.role + '\n<think>\n' + reasoning_content.strip('\n') + '\n</think>\n\n' + content.lstrip('\n') }}
+            {%- else %}
+                {{- '<|im_start|>' + message.role + '\n' + content }}
+            {%- endif %}
+        {%- else %}
+            {{- '<|im_start|>' + message.role + '\n' + content }}
+        {%- endif %}
+        {%- if message.tool_calls %}
+            {%- for tool_call in message.tool_calls %}
+                {%- if (loop.first and content) or (not loop.first) %}
+                    {{- '\n' }}
+                {%- endif %}
+                {%- if tool_call.function %}
+                    {%- set tool_call = tool_call.function %}
+                {%- endif %}
+                {{- '<tool_call>\n{"name": "' }}
+                {{- tool_call.name }}
+                {{- '", "arguments": ' }}
+                {%- if tool_call.arguments is string %}
+                    {{- tool_call.arguments }}
+                {%- else %}
+                    {{- tool_call.arguments | tojson }}
+                {%- endif %}
+                {{- '}\n</tool_call>' }}
+            {%- endfor %}
+        {%- endif %}
+        {{- '<|im_end|>\n' }}
+    {%- elif message.role == "tool" %}
+        {%- if loop.first or (messages[loop.index0 - 1].role != "tool") %}
+            {{- '<|im_start|>user' }}
+        {%- endif %}
+        {{- '\n<tool_response>\n' }}
+        {{- content }}
+        {{- '\n</tool_response>' }}
+        {%- if loop.last or (messages[loop.index0 + 1].role != "tool") %}
+            {{- '<|im_end|>\n' }}
+        {%- endif %}
+    {%- endif %}
+{%- endfor %}
+{%- if add_generation_prompt %}
+    {{- '<|im_start|>assistant\n' }}
+    {%- if enable_thinking is defined and enable_thinking is false %}
+        {{- '<think>\n\n</think>\n\n' }}
+    {%- endif %}
+{%- endif %}

Qwen3-32B/ll_bsearch_kl0.005_eap0.99_sha_bw5.53_4-5-6-7-8bit_grouped_seed42/config.json

@@ -0,0 +1,96 @@
+{
+  "architectures": [
+    "Qwen3ForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "bos_token_id": 151643,
+  "dtype": "float16",
+  "eos_token_id": 151645,
+  "head_dim": 128,
+  "hidden_act": "silu",
+  "hidden_size": 5120,
+  "initializer_range": 0.02,
+  "intermediate_size": 25600,
+  "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",
+    "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",
+    "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": 40960,
+  "max_window_layers": 64,
+  "model_type": "qwen3",
+  "num_attention_heads": 64,
+  "num_hidden_layers": 64,
+  "num_key_value_heads": 8,
+  "rms_norm_eps": 1e-06,
+  "rope_scaling": null,
+  "rope_theta": 1000000,
+  "sliding_window": null,
+  "tie_word_embeddings": false,
+  "transformers_version": "4.57.3",
+  "use_cache": false,
+  "use_sliding_window": false,
+  "vocab_size": 151936
+}

Qwen3-32B/ll_bsearch_kl0.005_eap0.99_sha_bw5.53_4-5-6-7-8bit_grouped_seed42/generation_config.json

@@ -0,0 +1,13 @@
+{
+  "bos_token_id": 151643,
+  "do_sample": true,
+  "eos_token_id": [
+    151645,
+    151643
+  ],
+  "pad_token_id": 151643,
+  "temperature": 0.6,
+  "top_k": 20,
+  "top_p": 0.95,
+  "transformers_version": "4.57.3"
+}

Qwen3-32B/ll_bsearch_kl0.005_eap0.99_sha_bw5.53_4-5-6-7-8bit_grouped_seed42/quantization_config.txt

@@ -0,0 +1,309 @@
+# Model: Qwen/Qwen3-32B
+# Layer directory: /nfs/scistore19/alistgrp/mhelcig/local/data/search/4_5_6_7_8bit_asym_g128/Qwen3-32B/4bit
+# Sensitivity method: shapley
+# Estimation method: permutation
+# Available bitwidths: [4, 5, 6, 7, 8]
+# Bitwidth map: {4: 4.156, 5: 5.156, 6: 6.156, 7: 7.156, 8: 8.156}
+#
+# Mode: binary_search_constraint (measured)
+# Constraint max_kl: 0.005
+# Constraint min_eap: 0.99
+# Weights: nll=0.0, kl=0.0, eap=1.0
+#
+# Average bitwidth: 5.5305
+# Total params: 19503513600
+# Total bits: 107863447962
+# Final KL: 0.001976
+# Final EAP: 0.990013
+# Final ETL: 0.009987
+# Satisfies constraints: True
+# Solver calls: 9
+# Evaluations: 9
+#
+# Bitwidth distribution:
+#   8-bit: 40 layers (14.3%)
+#   7-bit: 74 layers (26.4%)
+#   6-bit: 50 layers (17.9%)
+#   5-bit: 88 layers (31.4%)
+#   4-bit: 28 layers (10.0%)
+#
+model.layers.0.mlp.gate_proj: 4
+model.layers.0.mlp.up_proj: 4
+model.layers.0.self_attn.k_proj: 5
+model.layers.0.self_attn.q_proj: 5
+model.layers.0.self_attn.v_proj: 5
+model.layers.1.mlp.gate_proj: 5
+model.layers.1.mlp.up_proj: 5
+model.layers.1.self_attn.k_proj: 7
+model.layers.1.self_attn.q_proj: 7
+model.layers.1.self_attn.v_proj: 7
+model.layers.2.mlp.gate_proj: 4
+model.layers.2.mlp.up_proj: 4
+model.layers.2.self_attn.k_proj: 8
+model.layers.2.self_attn.q_proj: 8
+model.layers.2.self_attn.v_proj: 8
+model.layers.3.mlp.gate_proj: 4
+model.layers.3.mlp.up_proj: 4
+model.layers.3.self_attn.k_proj: 7
+model.layers.3.self_attn.q_proj: 7
+model.layers.3.self_attn.v_proj: 7
+model.layers.4.mlp.gate_proj: 4
+model.layers.4.mlp.up_proj: 4
+model.layers.4.self_attn.k_proj: 5
+model.layers.4.self_attn.q_proj: 5
+model.layers.4.self_attn.v_proj: 5
+model.layers.5.mlp.gate_proj: 5
+model.layers.5.mlp.up_proj: 5
+model.layers.5.self_attn.k_proj: 8
+model.layers.5.self_attn.q_proj: 8
+model.layers.5.self_attn.v_proj: 8
+model.layers.6.mlp.gate_proj: 4
+model.layers.6.mlp.up_proj: 4
+model.layers.6.self_attn.k_proj: 7
+model.layers.6.self_attn.q_proj: 7
+model.layers.6.self_attn.v_proj: 7
+model.layers.7.mlp.gate_proj: 5
+model.layers.7.mlp.up_proj: 5
+model.layers.7.self_attn.k_proj: 8
+model.layers.7.self_attn.q_proj: 8
+model.layers.7.self_attn.v_proj: 8
+model.layers.8.mlp.gate_proj: 4
+model.layers.8.mlp.up_proj: 4
+model.layers.8.self_attn.k_proj: 7
+model.layers.8.self_attn.q_proj: 7
+model.layers.8.self_attn.v_proj: 7
+model.layers.9.mlp.gate_proj: 5
+model.layers.9.mlp.up_proj: 5
+model.layers.9.self_attn.k_proj: 7
+model.layers.9.self_attn.q_proj: 7
+model.layers.9.self_attn.v_proj: 7
+model.layers.10.mlp.gate_proj: 5
+model.layers.10.mlp.up_proj: 5
+model.layers.10.self_attn.k_proj: 7
+model.layers.10.self_attn.q_proj: 7
+model.layers.10.self_attn.v_proj: 7
+model.layers.11.mlp.gate_proj: 4
+model.layers.11.mlp.up_proj: 4
+model.layers.11.self_attn.k_proj: 7
+model.layers.11.self_attn.q_proj: 7
+model.layers.11.self_attn.v_proj: 7
+model.layers.12.mlp.gate_proj: 5
+model.layers.12.mlp.up_proj: 5
+model.layers.12.self_attn.k_proj: 7
+model.layers.12.self_attn.q_proj: 7
+model.layers.12.self_attn.v_proj: 7
+model.layers.13.mlp.gate_proj: 5
+model.layers.13.mlp.up_proj: 5
+model.layers.13.self_attn.k_proj: 8
+model.layers.13.self_attn.q_proj: 8
+model.layers.13.self_attn.v_proj: 8
+model.layers.14.mlp.gate_proj: 5
+model.layers.14.mlp.up_proj: 5
+model.layers.14.self_attn.k_proj: 7
+model.layers.14.self_attn.q_proj: 7
+model.layers.14.self_attn.v_proj: 7
+model.layers.15.mlp.gate_proj: 5
+model.layers.15.mlp.up_proj: 5
+model.layers.15.self_attn.k_proj: 8
+model.layers.15.self_attn.q_proj: 8
+model.layers.15.self_attn.v_proj: 8
+model.layers.16.mlp.gate_proj: 5
+model.layers.16.mlp.up_proj: 5
+model.layers.16.self_attn.k_proj: 7
+model.layers.16.self_attn.q_proj: 7
+model.layers.16.self_attn.v_proj: 7
+model.layers.17.mlp.gate_proj: 5
+model.layers.17.mlp.up_proj: 5
+model.layers.17.self_attn.k_proj: 7
+model.layers.17.self_attn.q_proj: 7
+model.layers.17.self_attn.v_proj: 7
+model.layers.18.mlp.gate_proj: 5
+model.layers.18.mlp.up_proj: 5
+model.layers.18.self_attn.k_proj: 8
+model.layers.18.self_attn.q_proj: 8
+model.layers.18.self_attn.v_proj: 8
+model.layers.19.mlp.gate_proj: 5
+model.layers.19.mlp.up_proj: 5
+model.layers.19.self_attn.k_proj: 7
+model.layers.19.self_attn.q_proj: 7
+model.layers.19.self_attn.v_proj: 7
+model.layers.20.mlp.gate_proj: 4
+model.layers.20.mlp.up_proj: 4
+model.layers.20.self_attn.k_proj: 8
+model.layers.20.self_attn.q_proj: 8
+model.layers.20.self_attn.v_proj: 8
+model.layers.21.mlp.gate_proj: 5
+model.layers.21.mlp.up_proj: 5
+model.layers.21.self_attn.k_proj: 5
+model.layers.21.self_attn.q_proj: 5
+model.layers.21.self_attn.v_proj: 5
+model.layers.22.mlp.gate_proj: 5
+model.layers.22.mlp.up_proj: 5
+model.layers.22.self_attn.k_proj: 7
+model.layers.22.self_attn.q_proj: 7
+model.layers.22.self_attn.v_proj: 7
+model.layers.23.mlp.gate_proj: 6
+model.layers.23.mlp.up_proj: 6
+model.layers.23.self_attn.k_proj: 5
+model.layers.23.self_attn.q_proj: 5
+model.layers.23.self_attn.v_proj: 5
+model.layers.24.mlp.gate_proj: 5
+model.layers.24.mlp.up_proj: 5
+model.layers.24.self_attn.k_proj: 7
+model.layers.24.self_attn.q_proj: 7
+model.layers.24.self_attn.v_proj: 7
+model.layers.25.mlp.gate_proj: 5
+model.layers.25.mlp.up_proj: 5
+model.layers.25.self_attn.k_proj: 8
+model.layers.25.self_attn.q_proj: 8
+model.layers.25.self_attn.v_proj: 8
+model.layers.26.mlp.gate_proj: 5
+model.layers.26.mlp.up_proj: 5
+model.layers.26.self_attn.k_proj: 6
+model.layers.26.self_attn.q_proj: 6
+model.layers.26.self_attn.v_proj: 6
+model.layers.27.mlp.gate_proj: 5
+model.layers.27.mlp.up_proj: 5
+model.layers.27.self_attn.k_proj: 7
+model.layers.27.self_attn.q_proj: 7
+model.layers.27.self_attn.v_proj: 7
+model.layers.28.mlp.gate_proj: 5
+model.layers.28.mlp.up_proj: 5
+model.layers.28.self_attn.k_proj: 7
+model.layers.28.self_attn.q_proj: 7
+model.layers.28.self_attn.v_proj: 7
+model.layers.29.mlp.gate_proj: 6
+model.layers.29.mlp.up_proj: 6
+model.layers.29.self_attn.k_proj: 8
+model.layers.29.self_attn.q_proj: 8
+model.layers.29.self_attn.v_proj: 8
+model.layers.30.mlp.gate_proj: 6
+model.layers.30.mlp.up_proj: 6
+model.layers.30.self_attn.k_proj: 7
+model.layers.30.self_attn.q_proj: 7
+model.layers.30.self_attn.v_proj: 7
+model.layers.31.mlp.gate_proj: 6
+model.layers.31.mlp.up_proj: 6
+model.layers.31.self_attn.k_proj: 8
+model.layers.31.self_attn.q_proj: 8
+model.layers.31.self_attn.v_proj: 8
+model.layers.32.mlp.gate_proj: 6
+model.layers.32.mlp.up_proj: 6
+model.layers.32.self_attn.k_proj: 7
+model.layers.32.self_attn.q_proj: 7
+model.layers.32.self_attn.v_proj: 7
+model.layers.33.mlp.gate_proj: 6
+model.layers.33.mlp.up_proj: 6
+model.layers.33.self_attn.k_proj: 8
+model.layers.33.self_attn.q_proj: 8
+model.layers.33.self_attn.v_proj: 8
+model.layers.34.mlp.gate_proj: 6
+model.layers.34.mlp.up_proj: 6
+model.layers.34.self_attn.k_proj: 6
+model.layers.34.self_attn.q_proj: 6
+model.layers.34.self_attn.v_proj: 6
+model.layers.35.mlp.gate_proj: 5
+model.layers.35.mlp.up_proj: 5
+model.layers.35.self_attn.k_proj: 7
+model.layers.35.self_attn.q_proj: 7
+model.layers.35.self_attn.v_proj: 7
+model.layers.36.mlp.gate_proj: 7
+model.layers.36.mlp.up_proj: 7
+model.layers.36.self_attn.k_proj: 7
+model.layers.36.self_attn.q_proj: 7
+model.layers.36.self_attn.v_proj: 7
+model.layers.37.mlp.gate_proj: 6
+model.layers.37.mlp.up_proj: 6
+model.layers.37.self_attn.k_proj: 6
+model.layers.37.self_attn.q_proj: 6
+model.layers.37.self_attn.v_proj: 6
+model.layers.38.mlp.gate_proj: 6
+model.layers.38.mlp.up_proj: 6
+model.layers.38.self_attn.k_proj: 8
+model.layers.38.self_attn.q_proj: 8
+model.layers.38.self_attn.v_proj: 8
+model.layers.39.mlp.gate_proj: 7
+model.layers.39.mlp.up_proj: 7
+model.layers.39.self_attn.k_proj: 8
+model.layers.39.self_attn.q_proj: 8
+model.layers.39.self_attn.v_proj: 8
+model.layers.0.mlp.down_proj: 4
+model.layers.1.mlp.down_proj: 4
+model.layers.0.self_attn.o_proj: 5
+model.layers.2.self_attn.o_proj: 4
+model.layers.1.self_attn.o_proj: 5
+model.layers.29.mlp.down_proj: 7
+model.layers.21.mlp.down_proj: 6
+model.layers.3.self_attn.o_proj: 5
+model.layers.2.mlp.down_proj: 5
+model.layers.3.mlp.down_proj: 4
+model.layers.8.mlp.down_proj: 5
+model.layers.7.mlp.down_proj: 4
+model.layers.6.mlp.down_proj: 8
+model.layers.13.mlp.down_proj: 5
+model.layers.16.mlp.down_proj: 4
+model.layers.22.self_attn.o_proj: 6
+model.layers.38.mlp.down_proj: 6
+model.layers.16.self_attn.o_proj: 6
+model.layers.15.self_attn.o_proj: 5
+model.layers.7.self_attn.o_proj: 5
+model.layers.27.self_attn.o_proj: 5
+model.layers.28.mlp.down_proj: 6
+model.layers.5.self_attn.o_proj: 7
+model.layers.6.self_attn.o_proj: 7
+model.layers.12.mlp.down_proj: 5
+model.layers.23.self_attn.o_proj: 5
+model.layers.8.self_attn.o_proj: 6
+model.layers.19.self_attn.o_proj: 6
+model.layers.4.self_attn.o_proj: 6
+model.layers.4.mlp.down_proj: 5
+model.layers.5.mlp.down_proj: 4
+model.layers.9.mlp.down_proj: 4
+model.layers.9.self_attn.o_proj: 5
+model.layers.30.mlp.down_proj: 5
+model.layers.34.mlp.down_proj: 6
+model.layers.22.mlp.down_proj: 5
+model.layers.18.self_attn.o_proj: 6
+model.layers.20.self_attn.o_proj: 5
+model.layers.21.self_attn.o_proj: 5
+model.layers.38.self_attn.o_proj: 7
+model.layers.14.mlp.down_proj: 4
+model.layers.25.self_attn.o_proj: 6
+model.layers.11.mlp.down_proj: 5
+model.layers.10.self_attn.o_proj: 5
+model.layers.17.self_attn.o_proj: 4
+model.layers.34.self_attn.o_proj: 7
+model.layers.29.self_attn.o_proj: 5
+model.layers.39.self_attn.o_proj: 7
+model.layers.14.self_attn.o_proj: 5
+model.layers.20.mlp.down_proj: 5
+model.layers.32.mlp.down_proj: 6
+model.layers.37.self_attn.o_proj: 6
+model.layers.10.mlp.down_proj: 6
+model.layers.15.mlp.down_proj: 4
+model.layers.11.self_attn.o_proj: 7
+model.layers.12.self_attn.o_proj: 7
+model.layers.13.self_attn.o_proj: 6
+model.layers.18.mlp.down_proj: 6
+model.layers.31.mlp.down_proj: 5
+model.layers.32.self_attn.o_proj: 6
+model.layers.37.mlp.down_proj: 5
+model.layers.23.mlp.down_proj: 5
+model.layers.25.mlp.down_proj: 5
+model.layers.28.self_attn.o_proj: 6
+model.layers.24.mlp.down_proj: 5
+model.layers.17.mlp.down_proj: 5
+model.layers.19.mlp.down_proj: 4
+model.layers.26.self_attn.o_proj: 6
+model.layers.26.mlp.down_proj: 5
+model.layers.27.mlp.down_proj: 5
+model.layers.31.self_attn.o_proj: 6
+model.layers.24.self_attn.o_proj: 7
+model.layers.33.self_attn.o_proj: 6
+model.layers.30.self_attn.o_proj: 6
+model.layers.33.mlp.down_proj: 6
+model.layers.36.self_attn.o_proj: 7
+model.layers.39.mlp.down_proj: 5
+model.layers.35.mlp.down_proj: 5
+model.layers.36.mlp.down_proj: 5
+model.layers.35.self_attn.o_proj: 5

Qwen3-32B/ll_bsearch_kl0.005_eap0.99_sha_bw5.53_4-5-6-7-8bit_grouped_seed42/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": 131072,
+  "pad_token": "<|endoftext|>",
+  "split_special_tokens": false,
+  "tokenizer_class": "Qwen2Tokenizer",
+  "unk_token": null
+}