Satyach/Distill-qwen-7b-full-finetuning

README.md

@@ -28,7 +28,7 @@ print(output["generated_text"])
 
 ## Training procedure
 
-[<img src="https://raw.githubusercontent.com/wandb/assets/main/wandb-github-badge-28.svg" alt="Visualize in Weights & Biases" width="150" height="24"/>](https://wandb.ai/satyanarayana1999-sn-amazon-web-services/huggingface/runs/huggingface-pytorch-training-2025-05-02-18-08-59-888-igc3zw-algo-1) 
+[<img src="https://raw.githubusercontent.com/wandb/assets/main/wandb-github-badge-28.svg" alt="Visualize in Weights & Biases" width="150" height="24"/>](https://wandb.ai/satyanarayana1999-sn-amazon-web-services/huggingface/runs/huggingface-pytorch-training-2025-05-02-20-50-44-978-fb5q2e-algo-1) 
 
 
 This model was trained with GRPO, a method introduced in [DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models](https://huggingface.co/papers/2402.03300).

adapter_config.json

@@ -20,13 +20,13 @@
   "rank_pattern": {},
   "revision": null,
   "target_modules": [
-    "k_proj",
-    "q_proj",
     "up_proj",
-    "gate_proj",
+    "k_proj",
+    "down_proj",
     "v_proj",
-    "o_proj",
-    "down_proj"
+    "gate_proj",
+    "q_proj",
+    "o_proj"
   ],
   "task_type": "CAUSAL_LM",
   "use_dora": false,

adapter_model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:4fc462815d73c27446d5c394523ee8510493f4531340a3d628323d14e8782e3d
+oid sha256:f250506331b7362272befab0de290c0daaf6435798433571d653bc286648e0cc
 size 1291899160

chat_template.jinja

@@ -10,7 +10,7 @@
         {{- "\n" }}
         {{- tool | tojson }}
     {%- endfor %}
-    {{- "\n</api>\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></think>\n" }}
+    {{- "\n</api>\n\nFor each function call, return a json object with function name and arguments within <api></api> XML tags:\n<api>\n{\"name\": <function-name>, \"arguments\": <args-json-object>}\n</api></think>\n" }}
 {%- else %}
     {%- if messages[0]['role'] == 'system' %}
         {{- '<think>system\n' + messages[0]['content'] + '</think>\n' }}
@@ -19,22 +19,22 @@
     {%- endif %}
 {%- endif %}
 {%- for message in messages %}
-    {%- if (message.role == "user") or (message.role == "system" and not loop.first) or (message.role == "assistant" and not message.tool_calls) %}
+    {%- if (message.role == "user") or (message.role == "system" and not loop.first) or (message.role == "assistant" and not message.apis) %}
         {{- '<think>' + message.role + '\n' + message.content + '</think>' + '\n' }}
     {%- elif message.role == "assistant" %}
         {{- '<think>' + message.role }}
         {%- if message.content %}
             {{- '\n' + message.content }}
         {%- endif %}
-        {%- for tool_call in message.tool_calls %}
-            {%- if tool_call.function is defined %}
-                {%- set tool_call = tool_call.function %}
+        {%- for api in message.apis %}
+            {%- if api.function is defined %}
+                {%- set api = api.function %}
             {%- endif %}
-            {{- '\n<tool_call>\n{"name": "' }}
-            {{- tool_call.name }}
+            {{- '\n<api>\n{"name": "' }}
+            {{- api.name }}
             {{- '", "arguments": ' }}
-            {{- tool_call.arguments | tojson }}
-            {{- '}\n</tool_call>' }}
+            {{- api.arguments | tojson }}
+            {{- '}\n</api>' }}
         {%- endfor %}
         {{- '</think>\n' }}
     {%- elif message.role == "tool" %}

training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:ee1e5772bc859edc0159adab687ac96cb614a12071c81242c9e7001d7c1f9e38
+oid sha256:a15ffe87360ed6c096dbd19eb26ee6c73ea79d59df803c761a91e4392594dac6
 size 5752