Upload folder using huggingface_hub

infer/qwen/spider_data/sft__sft_qwen3_0.6b__ckpt1090__test__spider_data_sql_result.json

@@ -0,0 +1 @@
+[]

qwen3/sft_sft_qwen3_4b_spider_lora/e5-bs4-lr0.0001-G4-N2-NN1-lora-32-64-0.1/1090/README.md

@@ -1,9 +1,9 @@
 ---
-base_model: Qwen/Qwen3-4B
+base_model: Qwen/Qwen3-4B-Instruct-2507
 library_name: peft
 pipeline_tag: text-generation
 tags:
-- base_model:adapter:Qwen/Qwen3-4B
+- base_model:adapter:Qwen/Qwen3-4B-Instruct-2507
 - lora
 - transformers
 ---

qwen3/sft_sft_qwen3_4b_spider_lora/e5-bs4-lr0.0001-G4-N2-NN1-lora-32-64-0.1/1090/adapter_config.json

@@ -3,7 +3,7 @@
   "alpha_pattern": {},
   "arrow_config": null,
   "auto_mapping": null,
-  "base_model_name_or_path": "Qwen/Qwen3-4B",
+  "base_model_name_or_path": "Qwen/Qwen3-4B-Instruct-2507",
   "bias": "none",
   "corda_config": null,
   "ensure_weight_tying": false,
@@ -29,13 +29,13 @@
   "rank_pattern": {},
   "revision": null,
   "target_modules": [
-    "k_proj",
-    "q_proj",
-    "v_proj",
     "down_proj",
     "gate_proj",
-    "o_proj",
-    "up_proj"
+    "q_proj",
+    "up_proj",
+    "v_proj",
+    "k_proj",
+    "o_proj"
   ],
   "target_parameters": null,
   "task_type": "CAUSAL_LM",

qwen3/sft_sft_qwen3_4b_spider_lora/e5-bs4-lr0.0001-G4-N2-NN1-lora-32-64-0.1/1090/adapter_model.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:4f4bdd3a9d86cba1894d8d30bc49cbe40a353e240afbf5c210a594ea27c5b382
+oid sha256:6465f5c097501c718075fe7928288536990ef072eee9ac049b72c438e2d8277a
 size 132200851

qwen3/sft_sft_qwen3_4b_spider_lora/e5-bs4-lr0.0001-G4-N2-NN1-lora-32-64-0.1/1090/chat_template.jinja

@@ -14,14 +14,6 @@
         {{- '<|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 %}
@@ -31,24 +23,7 @@
     {%- 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 %}
+        {{- '<|im_start|>' + message.role + '\n' + content }}
         {%- if message.tool_calls %}
             {%- for tool_call in message.tool_calls %}
                 {%- if (loop.first and content) or (not loop.first) %}
@@ -83,7 +58,4 @@
 {%- 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/sft_sft_qwen3_4b_spider_lora/e5-bs4-lr0.0001-G4-N2-NN1-lora-32-64-0.1/1090/tokenizer_config.json

@@ -231,7 +231,7 @@
   "eos_token": "<|im_end|>",
   "errors": "replace",
   "extra_special_tokens": {},
-  "model_max_length": 131072,
+  "model_max_length": 1010000,
   "pad_token": "<|im_end|>",
   "padding_side": "right",
   "split_special_tokens": false,

qwen3/sft_sft_qwen3_4b_spider_lora/e5-bs4-lr0.0001-G4-N2-NN1-lora-32-64-0.1/218/README.md

@@ -1,9 +1,9 @@
 ---
-base_model: Qwen/Qwen3-4B
+base_model: Qwen/Qwen3-4B-Instruct-2507
 library_name: peft
 pipeline_tag: text-generation
 tags:
-- base_model:adapter:Qwen/Qwen3-4B
+- base_model:adapter:Qwen/Qwen3-4B-Instruct-2507
 - lora
 - transformers
 ---

qwen3/sft_sft_qwen3_4b_spider_lora/e5-bs4-lr0.0001-G4-N2-NN1-lora-32-64-0.1/218/adapter_config.json

@@ -3,7 +3,7 @@
   "alpha_pattern": {},
   "arrow_config": null,
   "auto_mapping": null,
-  "base_model_name_or_path": "Qwen/Qwen3-4B",
+  "base_model_name_or_path": "Qwen/Qwen3-4B-Instruct-2507",
   "bias": "none",
   "corda_config": null,
   "ensure_weight_tying": false,
@@ -29,13 +29,13 @@
   "rank_pattern": {},
   "revision": null,
   "target_modules": [
-    "k_proj",
-    "q_proj",
-    "v_proj",
     "down_proj",
     "gate_proj",
-    "o_proj",
-    "up_proj"
+    "q_proj",
+    "up_proj",
+    "v_proj",
+    "k_proj",
+    "o_proj"
   ],
   "target_parameters": null,
   "task_type": "CAUSAL_LM",

qwen3/sft_sft_qwen3_4b_spider_lora/e5-bs4-lr0.0001-G4-N2-NN1-lora-32-64-0.1/218/adapter_model.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:f56cb4af47a2d57a1e963a677c0e051922362f9b22c91187c58ac5fd77a17c57
+oid sha256:aa57bb9340e14e59218ab048a2febcee88a4b989afdfceae602cfa90d0888705
 size 132200851

qwen3/sft_sft_qwen3_4b_spider_lora/e5-bs4-lr0.0001-G4-N2-NN1-lora-32-64-0.1/218/chat_template.jinja

@@ -14,14 +14,6 @@
         {{- '<|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 %}
@@ -31,24 +23,7 @@
     {%- 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 %}
+        {{- '<|im_start|>' + message.role + '\n' + content }}
         {%- if message.tool_calls %}
             {%- for tool_call in message.tool_calls %}
                 {%- if (loop.first and content) or (not loop.first) %}
@@ -83,7 +58,4 @@
 {%- 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/sft_sft_qwen3_4b_spider_lora/e5-bs4-lr0.0001-G4-N2-NN1-lora-32-64-0.1/218/tokenizer_config.json

@@ -231,7 +231,7 @@
   "eos_token": "<|im_end|>",
   "errors": "replace",
   "extra_special_tokens": {},
-  "model_max_length": 131072,
+  "model_max_length": 1010000,
   "pad_token": "<|im_end|>",
   "padding_side": "right",
   "split_special_tokens": false,

qwen3/sft_sft_qwen3_4b_spider_lora/e5-bs4-lr0.0001-G4-N2-NN1-lora-32-64-0.1/436/README.md

@@ -1,9 +1,9 @@
 ---
-base_model: Qwen/Qwen3-4B
+base_model: Qwen/Qwen3-4B-Instruct-2507
 library_name: peft
 pipeline_tag: text-generation
 tags:
-- base_model:adapter:Qwen/Qwen3-4B
+- base_model:adapter:Qwen/Qwen3-4B-Instruct-2507
 - lora
 - transformers
 ---

qwen3/sft_sft_qwen3_4b_spider_lora/e5-bs4-lr0.0001-G4-N2-NN1-lora-32-64-0.1/436/adapter_config.json

@@ -3,7 +3,7 @@
   "alpha_pattern": {},
   "arrow_config": null,
   "auto_mapping": null,
-  "base_model_name_or_path": "Qwen/Qwen3-4B",
+  "base_model_name_or_path": "Qwen/Qwen3-4B-Instruct-2507",
   "bias": "none",
   "corda_config": null,
   "ensure_weight_tying": false,
@@ -29,13 +29,13 @@
   "rank_pattern": {},
   "revision": null,
   "target_modules": [
-    "k_proj",
-    "q_proj",
-    "v_proj",
     "down_proj",
     "gate_proj",
-    "o_proj",
-    "up_proj"
+    "q_proj",
+    "up_proj",
+    "v_proj",
+    "k_proj",
+    "o_proj"
   ],
   "target_parameters": null,
   "task_type": "CAUSAL_LM",

qwen3/sft_sft_qwen3_4b_spider_lora/e5-bs4-lr0.0001-G4-N2-NN1-lora-32-64-0.1/436/adapter_model.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:21208063ded5c5c886e2cd41bcf3137351ee6ea1bf98440e52e4d0bcc46cd9a4
+oid sha256:62fa33594d3a4d460651c7bbc4edfacf2582d74c1b2247d3b14bbcb32a4d34d4
 size 132200851

qwen3/sft_sft_qwen3_4b_spider_lora/e5-bs4-lr0.0001-G4-N2-NN1-lora-32-64-0.1/436/chat_template.jinja

@@ -14,14 +14,6 @@
         {{- '<|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 %}
@@ -31,24 +23,7 @@
     {%- 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 %}
+        {{- '<|im_start|>' + message.role + '\n' + content }}
         {%- if message.tool_calls %}
             {%- for tool_call in message.tool_calls %}
                 {%- if (loop.first and content) or (not loop.first) %}
@@ -83,7 +58,4 @@
 {%- 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/sft_sft_qwen3_4b_spider_lora/e5-bs4-lr0.0001-G4-N2-NN1-lora-32-64-0.1/436/tokenizer_config.json

@@ -231,7 +231,7 @@
   "eos_token": "<|im_end|>",
   "errors": "replace",
   "extra_special_tokens": {},
-  "model_max_length": 131072,
+  "model_max_length": 1010000,
   "pad_token": "<|im_end|>",
   "padding_side": "right",
   "split_special_tokens": false,

qwen3/sft_sft_qwen3_4b_spider_lora/e5-bs4-lr0.0001-G4-N2-NN1-lora-32-64-0.1/654/README.md

@@ -1,9 +1,9 @@
 ---
-base_model: Qwen/Qwen3-4B
+base_model: Qwen/Qwen3-4B-Instruct-2507
 library_name: peft
 pipeline_tag: text-generation
 tags:
-- base_model:adapter:Qwen/Qwen3-4B
+- base_model:adapter:Qwen/Qwen3-4B-Instruct-2507
 - lora
 - transformers
 ---

qwen3/sft_sft_qwen3_4b_spider_lora/e5-bs4-lr0.0001-G4-N2-NN1-lora-32-64-0.1/654/adapter_config.json

@@ -3,7 +3,7 @@
   "alpha_pattern": {},
   "arrow_config": null,
   "auto_mapping": null,
-  "base_model_name_or_path": "Qwen/Qwen3-4B",
+  "base_model_name_or_path": "Qwen/Qwen3-4B-Instruct-2507",
   "bias": "none",
   "corda_config": null,
   "ensure_weight_tying": false,
@@ -29,13 +29,13 @@
   "rank_pattern": {},
   "revision": null,
   "target_modules": [
-    "k_proj",
-    "q_proj",
-    "v_proj",
     "down_proj",
     "gate_proj",
-    "o_proj",
-    "up_proj"
+    "q_proj",
+    "up_proj",
+    "v_proj",
+    "k_proj",
+    "o_proj"
   ],
   "target_parameters": null,
   "task_type": "CAUSAL_LM",

qwen3/sft_sft_qwen3_4b_spider_lora/e5-bs4-lr0.0001-G4-N2-NN1-lora-32-64-0.1/654/adapter_model.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:bf85f9f3b57ae4d77c196ef2a5c9c33031dc6b77025767be6d0b1c024cfd3b24
+oid sha256:95cfd0f09243e85433b6fb67eeb39699df8dfacb808c119b24042cd0f3a2c6bf
 size 132200851

qwen3/sft_sft_qwen3_4b_spider_lora/e5-bs4-lr0.0001-G4-N2-NN1-lora-32-64-0.1/654/chat_template.jinja

@@ -14,14 +14,6 @@
         {{- '<|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 %}
@@ -31,24 +23,7 @@
     {%- 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 %}
+        {{- '<|im_start|>' + message.role + '\n' + content }}
         {%- if message.tool_calls %}
             {%- for tool_call in message.tool_calls %}
                 {%- if (loop.first and content) or (not loop.first) %}
@@ -83,7 +58,4 @@
 {%- 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/sft_sft_qwen3_4b_spider_lora/e5-bs4-lr0.0001-G4-N2-NN1-lora-32-64-0.1/654/tokenizer_config.json

@@ -231,7 +231,7 @@
   "eos_token": "<|im_end|>",
   "errors": "replace",
   "extra_special_tokens": {},
-  "model_max_length": 131072,
+  "model_max_length": 1010000,
   "pad_token": "<|im_end|>",
   "padding_side": "right",
   "split_special_tokens": false,

qwen3/sft_sft_qwen3_4b_spider_lora/e5-bs4-lr0.0001-G4-N2-NN1-lora-32-64-0.1/872/README.md

@@ -1,9 +1,9 @@
 ---
-base_model: Qwen/Qwen3-4B
+base_model: Qwen/Qwen3-4B-Instruct-2507
 library_name: peft
 pipeline_tag: text-generation
 tags:
-- base_model:adapter:Qwen/Qwen3-4B
+- base_model:adapter:Qwen/Qwen3-4B-Instruct-2507
 - lora
 - transformers
 ---

qwen3/sft_sft_qwen3_4b_spider_lora/e5-bs4-lr0.0001-G4-N2-NN1-lora-32-64-0.1/872/adapter_config.json

@@ -3,7 +3,7 @@
   "alpha_pattern": {},
   "arrow_config": null,
   "auto_mapping": null,
-  "base_model_name_or_path": "Qwen/Qwen3-4B",
+  "base_model_name_or_path": "Qwen/Qwen3-4B-Instruct-2507",
   "bias": "none",
   "corda_config": null,
   "ensure_weight_tying": false,
@@ -29,13 +29,13 @@
   "rank_pattern": {},
   "revision": null,
   "target_modules": [
-    "k_proj",
-    "q_proj",
-    "v_proj",
     "down_proj",
     "gate_proj",
-    "o_proj",
-    "up_proj"
+    "q_proj",
+    "up_proj",
+    "v_proj",
+    "k_proj",
+    "o_proj"
   ],
   "target_parameters": null,
   "task_type": "CAUSAL_LM",

qwen3/sft_sft_qwen3_4b_spider_lora/e5-bs4-lr0.0001-G4-N2-NN1-lora-32-64-0.1/872/adapter_model.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:2e627c79b54162cebf93f1bfa52f3e29c504bacdadbc4e1aa2e5aeff46b4b2f3
+oid sha256:24f1f070ca1f20cea2b2d4b9d0ab1ebcc42dc23f6e84e42700dea2fc20828a69
 size 132200851

qwen3/sft_sft_qwen3_4b_spider_lora/e5-bs4-lr0.0001-G4-N2-NN1-lora-32-64-0.1/872/chat_template.jinja

@@ -14,14 +14,6 @@
         {{- '<|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 %}
@@ -31,24 +23,7 @@
     {%- 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 %}
+        {{- '<|im_start|>' + message.role + '\n' + content }}
         {%- if message.tool_calls %}
             {%- for tool_call in message.tool_calls %}
                 {%- if (loop.first and content) or (not loop.first) %}
@@ -83,7 +58,4 @@
 {%- 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/sft_sft_qwen3_4b_spider_lora/e5-bs4-lr0.0001-G4-N2-NN1-lora-32-64-0.1/872/tokenizer_config.json

@@ -231,7 +231,7 @@
   "eos_token": "<|im_end|>",
   "errors": "replace",
   "extra_special_tokens": {},
-  "model_max_length": 131072,
+  "model_max_length": 1010000,
   "pad_token": "<|im_end|>",
   "padding_side": "right",
   "split_special_tokens": false,

qwen3/sft_sft_qwen3_4b_spider_lora/e5-bs4-lr0.0001-G4-N2-NN1-lora-32-64-0.1/args.json

@@ -1 +1 @@
-{"model_path": "Qwen/Qwen3-4B", "ckpt_name": "qwen3-4B", "model_type": "qwen", "teacher_model_type": null, "n_gpu": 2, "n_nodes": 1, "teacher_model_path": null, "teacher_ckpt_name": null, "teacher_model_fp16": false, "model_parallel": false, "model_parallel_size": null, "no_value": false, "dropout_path_rate": null, "fp32": false, "bf16": false, "type": "lm", "do_train": true, "do_valid": true, "do_eval": false, "base_path": ".", "load": null, "save": "./results/qwen3/sft_sft_qwen3_4b_spider_lora/e5-bs4-lr0.0001-G4-N2-NN1-lora-32-64-0.1", "log_interval": 20, "mid_log_num": -1, "save_interval": -1, "eval_interval": -1, "local_rank": 0, "save_additional_suffix": "", "save_rollout": false, "eb_sample_times": 3, "split": null, "data_dir": "processed_data/benchmarks/spider_data/qwen", "processed_data_dir": null, "force_process": false, "force_process_demo": false, "data_process_workers": -1, "train_num": -1, "train_ratio": 1, "dev_num": -1, "dev_ratio": 1, "gen_num": -1, "data_names": null, "prompt_type": null, "num_workers": 0, "max_prompt_length": 1479, "t_max_prompt_length": 640, "min_prompt_length": 128, "json_data": false, "bin_data": false, "txt_data": false, "prompt_data_dir": null, "lm_data_dir": null, "eval_ppl": false, "eval_rw": false, "eval_gen": true, "only_prompt": false, "slice_data": false, "batch_size": 4, "eval_batch_size": 16, "clip_grad": 1.0, "total_iters": null, "train_iters_per_epoch": -1, "max_length": 1612, "t_max_length": 1024, "seed": 42, "seed_order": 42, "seed_data": 42, "seed_ppo": 42, "seed_lm": 7, "epochs": 5, "training_epochs": 10000, "gradient_accumulation_steps": 4, "gradient_checkpointing": true, "attn_dtype": null, "lr": 0.0001, "lr_min": 1e-07, "weight_decay": 0.01, "loss_scale": 65536, "kd_ratio": null, "warmup_iters": 0, "warmup_ratio": 0.1, "lr_decay_iters": null, "lr_decay_style": "wrmup_cosine", "scheduler_name": "constant_trm", "w_span_loss": 1.0, "reward_scaling": null, "cliprange_reward": 1, "ppo_epochs": null, "num_rollouts": 256, "num_rollouts_per_device": null, "cliprange": 0.2, "chunk_size": null, "gamma": 0.95, "length_norm": false, "single_step_reg": false, "teacher_mixed_alpha": null, "lm_coef": 1, "skew_alpha": 0.1, "student_gen": false, "gen_top_p": 1.0, "gen_num_beams": 2, "mixed_alpha": 0.5, "loss_eps": 0.1, "init_threshold": 0.0, "capacity": 1000, "replay_ratio": "decreasing", "student_layer_mapping": [-1], "teacher_layer_mapping": [-1], "split_layer_mapping": [0, 0, 0, 0], "use_dsa": false, "use_hs": false, "top_k": 0, "top_p": 0.95, "do_sample": true, "no_repeat_ngram_size": 6, "repetition_penalty": null, "num_beams": 1, "temperature": 0.5, "peft": "lora", "peft_lora_r": 32, "peft_lora_alpha": 64, "peft_lora_dropout": 0.1, "peft_name": null, "peft_path": null, "teacher_peft_name": null, "teacher_peft_path": null, "deepspeed": true, "deepspeed_config": "./configs/deepspeed/ds_config_bf16.json", "deepscale": false, "deepscale_config": null, "rank": 0, "world_size": 2}
+{"model_path": "Qwen/Qwen3-4B-Instruct-2507", "ckpt_name": "qwen3-4B", "model_type": "qwen", "teacher_model_type": null, "n_gpu": 2, "n_nodes": 1, "teacher_model_path": null, "teacher_ckpt_name": null, "teacher_model_fp16": false, "model_parallel": false, "model_parallel_size": null, "no_value": false, "dropout_path_rate": null, "fp32": false, "bf16": false, "type": "lm", "do_train": true, "do_valid": true, "do_eval": false, "base_path": ".", "load": null, "save": "./results/qwen3/sft_sft_qwen3_4b_spider_lora/e5-bs4-lr0.0001-G4-N2-NN1-lora-32-64-0.1", "log_interval": 20, "mid_log_num": -1, "save_interval": -1, "eval_interval": -1, "local_rank": 0, "save_additional_suffix": "", "save_rollout": false, "eb_sample_times": 3, "split": null, "data_dir": "processed_data/benchmarks/spider_data/qwen", "processed_data_dir": null, "force_process": false, "force_process_demo": false, "data_process_workers": -1, "train_num": -1, "train_ratio": 1, "dev_num": -1, "dev_ratio": 1, "gen_num": -1, "data_names": null, "prompt_type": null, "num_workers": 0, "max_prompt_length": 1479, "t_max_prompt_length": 640, "min_prompt_length": 128, "json_data": false, "bin_data": false, "txt_data": false, "prompt_data_dir": null, "lm_data_dir": null, "eval_ppl": false, "eval_rw": false, "eval_gen": true, "only_prompt": false, "slice_data": false, "batch_size": 4, "eval_batch_size": 16, "clip_grad": 1.0, "total_iters": null, "train_iters_per_epoch": -1, "max_length": 1612, "t_max_length": 1024, "seed": 42, "seed_order": 42, "seed_data": 42, "seed_ppo": 42, "seed_lm": 7, "epochs": 5, "training_epochs": 10000, "gradient_accumulation_steps": 4, "gradient_checkpointing": true, "attn_dtype": null, "lr": 0.0001, "lr_min": 1e-07, "weight_decay": 0.01, "loss_scale": 65536, "kd_ratio": null, "warmup_iters": 0, "warmup_ratio": 0.1, "lr_decay_iters": null, "lr_decay_style": "wrmup_cosine", "scheduler_name": "constant_trm", "w_span_loss": 1.0, "reward_scaling": null, "cliprange_reward": 1, "ppo_epochs": null, "num_rollouts": 256, "num_rollouts_per_device": null, "cliprange": 0.2, "chunk_size": null, "gamma": 0.95, "length_norm": false, "single_step_reg": false, "teacher_mixed_alpha": null, "lm_coef": 1, "skew_alpha": 0.1, "student_gen": false, "gen_top_p": 1.0, "gen_num_beams": 2, "mixed_alpha": 0.5, "loss_eps": 0.1, "init_threshold": 0.0, "capacity": 1000, "replay_ratio": "decreasing", "student_layer_mapping": [-1], "teacher_layer_mapping": [-1], "split_layer_mapping": [0, 0, 0, 0], "use_dsa": false, "use_hs": false, "top_k": 0, "top_p": 0.95, "do_sample": true, "no_repeat_ngram_size": 6, "repetition_penalty": null, "num_beams": 1, "temperature": 0.5, "peft": "lora", "peft_lora_r": 32, "peft_lora_alpha": 64, "peft_lora_dropout": 0.1, "peft_name": null, "peft_path": null, "teacher_peft_name": null, "teacher_peft_path": null, "deepspeed": true, "deepspeed_config": "./configs/deepspeed/ds_config_bf16.json", "deepscale": false, "deepscale_config": null, "rank": 0, "world_size": 2}

qwen3/sft_sft_qwen3_4b_spider_lora/e5-bs4-lr0.0001-G4-N2-NN1-lora-32-64-0.1/log.txt

@@ -61,3 +61,66 @@ train | epoch   4 | Iter:   4156/  4360 | global iter:   1040/  1090 | loss: 0.0
 train | epoch   4 | Iter:   4236/  4360 | global iter:   1060/  1090 | loss: 0.0028 | ds_loss: 0.0000 | lr: 2.4619e-07 | scale:     1.0000 | micro time: 5.292 | step time: 21.087
 train | epoch   4 | Iter:   4316/  4360 | global iter:   1080/  1090 | loss: 0.0036 | ds_loss: 0.0000 | lr: 3.1020e-08 | scale:     1.0000 | micro time: 5.287 | step time: 21.087
 dev | avg_loss: 0.4173325047348485 | {'exact_match': 52.9981, 'rougeL': 88.8055}
+
+
+============================== EXP at 2026-06-12 05:29:11 ==============================
+dev | avg_loss: 2.152462121212121 | {'exact_match': 21.0832, 'rougeL': 71.8971}
+train | epoch   0 | Iter:     76/  4360 | global iter:     20/  1090 | loss: 1.6727 | ds_loss: 0.0000 | lr: 1.7431e-05 | scale:     1.0000 | micro time: 5.289 | step time: 20.024
+train | epoch   0 | Iter:    156/  4360 | global iter:     40/  1090 | loss: 0.4101 | ds_loss: 0.0000 | lr: 3.5780e-05 | scale:     1.0000 | micro time: 5.287 | step time: 21.078
+train | epoch   0 | Iter:    236/  4360 | global iter:     60/  1090 | loss: 0.2133 | ds_loss: 0.0000 | lr: 5.4128e-05 | scale:     1.0000 | micro time: 5.288 | step time: 21.079
+train | epoch   0 | Iter:    316/  4360 | global iter:     80/  1090 | loss: 0.1717 | ds_loss: 0.0000 | lr: 7.2477e-05 | scale:     1.0000 | micro time: 5.281 | step time: 21.078
+train | epoch   0 | Iter:    396/  4360 | global iter:    100/  1090 | loss: 0.1598 | ds_loss: 0.0000 | lr: 9.0826e-05 | scale:     1.0000 | micro time: 5.286 | step time: 21.081
+train | epoch   0 | Iter:    476/  4360 | global iter:    120/  1090 | loss: 0.1446 | ds_loss: 0.0000 | lr: 9.9974e-05 | scale:     1.0000 | micro time: 5.283 | step time: 21.080
+train | epoch   0 | Iter:    556/  4360 | global iter:    140/  1090 | loss: 0.1296 | ds_loss: 0.0000 | lr: 9.9769e-05 | scale:     1.0000 | micro time: 5.284 | step time: 21.081
+train | epoch   0 | Iter:    636/  4360 | global iter:    160/  1090 | loss: 0.1280 | ds_loss: 0.0000 | lr: 9.9360e-05 | scale:     1.0000 | micro time: 5.284 | step time: 21.076
+train | epoch   0 | Iter:    716/  4360 | global iter:    180/  1090 | loss: 0.1159 | ds_loss: 0.0000 | lr: 9.8749e-05 | scale:     1.0000 | micro time: 5.285 | step time: 21.081
+train | epoch   0 | Iter:    796/  4360 | global iter:    200/  1090 | loss: 0.1104 | ds_loss: 0.0000 | lr: 9.7938e-05 | scale:     1.0000 | micro time: 5.283 | step time: 21.077
+dev | avg_loss: 0.19998816287878787 | {'exact_match': 53.7718, 'rougeL': 88.4715}
+train | epoch   1 | Iter:    876/  4360 | global iter:    220/  1090 | loss: 0.1118 | ds_loss: 0.0000 | lr: 9.6930e-05 | scale:     1.0000 | micro time: 5.281 | step time: 21.079
+train | epoch   1 | Iter:    956/  4360 | global iter:    240/  1090 | loss: 0.0764 | ds_loss: 0.0000 | lr: 9.5729e-05 | scale:     1.0000 | micro time: 5.282 | step time: 21.082
+train | epoch   1 | Iter:   1036/  4360 | global iter:    260/  1090 | loss: 0.0793 | ds_loss: 0.0000 | lr: 9.4341e-05 | scale:     1.0000 | micro time: 5.288 | step time: 21.082
+train | epoch   1 | Iter:   1116/  4360 | global iter:    280/  1090 | loss: 0.0756 | ds_loss: 0.0000 | lr: 9.2772e-05 | scale:     1.0000 | micro time: 5.290 | step time: 21.085
+train | epoch   1 | Iter:   1196/  4360 | global iter:    300/  1090 | loss: 0.0735 | ds_loss: 0.0000 | lr: 9.1026e-05 | scale:     1.0000 | micro time: 5.285 | step time: 21.085
+train | epoch   1 | Iter:   1276/  4360 | global iter:    320/  1090 | loss: 0.0662 | ds_loss: 0.0000 | lr: 8.9113e-05 | scale:     1.0000 | micro time: 5.287 | step time: 21.080
+train | epoch   1 | Iter:   1356/  4360 | global iter:    340/  1090 | loss: 0.0659 | ds_loss: 0.0000 | lr: 8.7039e-05 | scale:     1.0000 | micro time: 5.291 | step time: 21.081
+train | epoch   1 | Iter:   1436/  4360 | global iter:    360/  1090 | loss: 0.0668 | ds_loss: 0.0000 | lr: 8.4813e-05 | scale:     1.0000 | micro time: 5.290 | step time: 21.085
+train | epoch   1 | Iter:   1516/  4360 | global iter:    380/  1090 | loss: 0.0683 | ds_loss: 0.0000 | lr: 8.2445e-05 | scale:     1.0000 | micro time: 5.290 | step time: 21.083
+train | epoch   1 | Iter:   1596/  4360 | global iter:    400/  1090 | loss: 0.0688 | ds_loss: 0.0000 | lr: 7.9943e-05 | scale:     1.0000 | micro time: 5.288 | step time: 21.082
+train | epoch   1 | Iter:   1676/  4360 | global iter:    420/  1090 | loss: 0.0608 | ds_loss: 0.0000 | lr: 7.7319e-05 | scale:     1.0000 | micro time: 5.288 | step time: 21.085
+dev | avg_loss: 0.23133433948863635 | {'exact_match': 52.7079, 'rougeL': 88.6921}
+train | epoch   2 | Iter:   1756/  4360 | global iter:    440/  1090 | loss: 0.0584 | ds_loss: 0.0000 | lr: 7.4583e-05 | scale:     1.0000 | micro time: 5.288 | step time: 21.081
+train | epoch   2 | Iter:   1836/  4360 | global iter:    460/  1090 | loss: 0.0332 | ds_loss: 0.0000 | lr: 7.1746e-05 | scale:     1.0000 | micro time: 5.283 | step time: 21.079
+train | epoch   2 | Iter:   1916/  4360 | global iter:    480/  1090 | loss: 0.0367 | ds_loss: 0.0000 | lr: 6.8819e-05 | scale:     1.0000 | micro time: 5.282 | step time: 21.081
+train | epoch   2 | Iter:   1996/  4360 | global iter:    500/  1090 | loss: 0.0354 | ds_loss: 0.0000 | lr: 6.5816e-05 | scale:     1.0000 | micro time: 5.285 | step time: 21.081
+train | epoch   2 | Iter:   2076/  4360 | global iter:    520/  1090 | loss: 0.0334 | ds_loss: 0.0000 | lr: 6.2748e-05 | scale:     1.0000 | micro time: 5.287 | step time: 21.078
+train | epoch   2 | Iter:   2156/  4360 | global iter:    540/  1090 | loss: 0.0299 | ds_loss: 0.0000 | lr: 5.9627e-05 | scale:     1.0000 | micro time: 5.289 | step time: 21.079
+train | epoch   2 | Iter:   2236/  4360 | global iter:    560/  1090 | loss: 0.0307 | ds_loss: 0.0000 | lr: 5.6467e-05 | scale:     1.0000 | micro time: 5.285 | step time: 21.084
+train | epoch   2 | Iter:   2316/  4360 | global iter:    580/  1090 | loss: 0.0303 | ds_loss: 0.0000 | lr: 5.3280e-05 | scale:     1.0000 | micro time: 5.281 | step time: 21.079
+train | epoch   2 | Iter:   2396/  4360 | global iter:    600/  1090 | loss: 0.0244 | ds_loss: 0.0000 | lr: 5.0080e-05 | scale:     1.0000 | micro time: 5.281 | step time: 21.080
+train | epoch   2 | Iter:   2476/  4360 | global iter:    620/  1090 | loss: 0.0311 | ds_loss: 0.0000 | lr: 4.6880e-05 | scale:     1.0000 | micro time: 5.285 | step time: 21.081
+train | epoch   2 | Iter:   2556/  4360 | global iter:    640/  1090 | loss: 0.0267 | ds_loss: 0.0000 | lr: 4.3692e-05 | scale:     1.0000 | micro time: 5.283 | step time: 21.079
+dev | avg_loss: 0.2874348958333333 | {'exact_match': 52.0309, 'rougeL': 88.7671}
+train | epoch   3 | Iter:   2636/  4360 | global iter:    660/  1090 | loss: 0.0236 | ds_loss: 0.0000 | lr: 4.0530e-05 | scale:     1.0000 | micro time: 5.284 | step time: 21.080
+train | epoch   3 | Iter:   2716/  4360 | global iter:    680/  1090 | loss: 0.0128 | ds_loss: 0.0000 | lr: 3.7407e-05 | scale:     1.0000 | micro time: 5.289 | step time: 21.086
+train | epoch   3 | Iter:   2796/  4360 | global iter:    700/  1090 | loss: 0.0114 | ds_loss: 0.0000 | lr: 3.4336e-05 | scale:     1.0000 | micro time: 5.288 | step time: 21.082
+train | epoch   3 | Iter:   2876/  4360 | global iter:    720/  1090 | loss: 0.0114 | ds_loss: 0.0000 | lr: 3.1329e-05 | scale:     1.0000 | micro time: 5.287 | step time: 21.085
+train | epoch   3 | Iter:   2956/  4360 | global iter:    740/  1090 | loss: 0.0101 | ds_loss: 0.0000 | lr: 2.8399e-05 | scale:     1.0000 | micro time: 5.285 | step time: 21.082
+train | epoch   3 | Iter:   3036/  4360 | global iter:    760/  1090 | loss: 0.0110 | ds_loss: 0.0000 | lr: 2.5557e-05 | scale:     1.0000 | micro time: 5.285 | step time: 21.084
+train | epoch   3 | Iter:   3116/  4360 | global iter:    780/  1090 | loss: 0.0100 | ds_loss: 0.0000 | lr: 2.2815e-05 | scale:     1.0000 | micro time: 5.283 | step time: 21.084
+train | epoch   3 | Iter:   3196/  4360 | global iter:    800/  1090 | loss: 0.0103 | ds_loss: 0.0000 | lr: 2.0185e-05 | scale:     1.0000 | micro time: 5.290 | step time: 21.084
+train | epoch   3 | Iter:   3276/  4360 | global iter:    820/  1090 | loss: 0.0093 | ds_loss: 0.0000 | lr: 1.7677e-05 | scale:     1.0000 | micro time: 5.291 | step time: 21.085
+train | epoch   3 | Iter:   3356/  4360 | global iter:    840/  1090 | loss: 0.0087 | ds_loss: 0.0000 | lr: 1.5302e-05 | scale:     1.0000 | micro time: 5.285 | step time: 21.085
+train | epoch   3 | Iter:   3436/  4360 | global iter:    860/  1090 | loss: 0.0096 | ds_loss: 0.0000 | lr: 1.3069e-05 | scale:     1.0000 | micro time: 5.287 | step time: 21.087
+dev | avg_loss: 0.3450520833333333 | {'exact_match': 52.6112, 'rougeL': 88.9668}
+train | epoch   4 | Iter:   3516/  4360 | global iter:    880/  1090 | loss: 0.0077 | ds_loss: 0.0000 | lr: 1.0987e-05 | scale:     1.0000 | micro time: 5.285 | step time: 21.081
+train | epoch   4 | Iter:   3596/  4360 | global iter:    900/  1090 | loss: 0.0041 | ds_loss: 0.0000 | lr: 9.0654e-06 | scale:     1.0000 | micro time: 5.286 | step time: 21.084
+train | epoch   4 | Iter:   3676/  4360 | global iter:    920/  1090 | loss: 0.0029 | ds_loss: 0.0000 | lr: 7.3116e-06 | scale:     1.0000 | micro time: 5.287 | step time: 21.084
+train | epoch   4 | Iter:   3756/  4360 | global iter:    940/  1090 | loss: 0.0039 | ds_loss: 0.0000 | lr: 5.7329e-06 | scale:     1.0000 | micro time: 5.285 | step time: 21.084
+train | epoch   4 | Iter:   3836/  4360 | global iter:    960/  1090 | loss: 0.0035 | ds_loss: 0.0000 | lr: 4.3358e-06 | scale:     1.0000 | micro time: 5.282 | step time: 21.085
+train | epoch   4 | Iter:   3916/  4360 | global iter:    980/  1090 | loss: 0.0030 | ds_loss: 0.0000 | lr: 3.1259e-06 | scale:     1.0000 | micro time: 5.293 | step time: 21.083
+train | epoch   4 | Iter:   3996/  4360 | global iter:   1000/  1090 | loss: 0.0045 | ds_loss: 0.0000 | lr: 2.1082e-06 | scale:     1.0000 | micro time: 5.286 | step time: 21.088
+train | epoch   4 | Iter:   4076/  4360 | global iter:   1020/  1090 | loss: 0.0038 | ds_loss: 0.0000 | lr: 1.2869e-06 | scale:     1.0000 | micro time: 5.291 | step time: 21.085
+train | epoch   4 | Iter:   4156/  4360 | global iter:   1040/  1090 | loss: 0.0029 | ds_loss: 0.0000 | lr: 6.6539e-07 | scale:     1.0000 | micro time: 5.286 | step time: 21.085
+train | epoch   4 | Iter:   4236/  4360 | global iter:   1060/  1090 | loss: 0.0029 | ds_loss: 0.0000 | lr: 2.4619e-07 | scale:     1.0000 | micro time: 5.285 | step time: 21.083
+train | epoch   4 | Iter:   4316/  4360 | global iter:   1080/  1090 | loss: 0.0044 | ds_loss: 0.0000 | lr: 3.1020e-08 | scale:     1.0000 | micro time: 5.289 | step time: 21.087
+dev | avg_loss: 0.38920454545454547 | {'exact_match': 54.1586, 'rougeL': 89.1636}