Super-squash branch 'main' using huggingface_hub

.gitattributes

@@ -0,0 +1,38 @@
+*.7z filter=lfs diff=lfs merge=lfs -text
+*.arrow filter=lfs diff=lfs merge=lfs -text
+*.bin filter=lfs diff=lfs merge=lfs -text
+*.bz2 filter=lfs diff=lfs merge=lfs -text
+*.ckpt filter=lfs diff=lfs merge=lfs -text
+*.ftz filter=lfs diff=lfs merge=lfs -text
+*.gz filter=lfs diff=lfs merge=lfs -text
+*.h5 filter=lfs diff=lfs merge=lfs -text
+*.joblib filter=lfs diff=lfs merge=lfs -text
+*.lfs.* filter=lfs diff=lfs merge=lfs -text
+*.mlmodel filter=lfs diff=lfs merge=lfs -text
+*.model filter=lfs diff=lfs merge=lfs -text
+*.msgpack filter=lfs diff=lfs merge=lfs -text
+*.npy filter=lfs diff=lfs merge=lfs -text
+*.npz filter=lfs diff=lfs merge=lfs -text
+*.onnx filter=lfs diff=lfs merge=lfs -text
+*.ot filter=lfs diff=lfs merge=lfs -text
+*.parquet filter=lfs diff=lfs merge=lfs -text
+*.pb filter=lfs diff=lfs merge=lfs -text
+*.pickle filter=lfs diff=lfs merge=lfs -text
+*.pkl filter=lfs diff=lfs merge=lfs -text
+*.pt filter=lfs diff=lfs merge=lfs -text
+*.pth filter=lfs diff=lfs merge=lfs -text
+*.rar filter=lfs diff=lfs merge=lfs -text
+*.safetensors filter=lfs diff=lfs merge=lfs -text
+saved_model/**/* filter=lfs diff=lfs merge=lfs -text
+*.tar.* filter=lfs diff=lfs merge=lfs -text
+*.tar filter=lfs diff=lfs merge=lfs -text
+*.tflite filter=lfs diff=lfs merge=lfs -text
+*.tgz filter=lfs diff=lfs merge=lfs -text
+*.wasm filter=lfs diff=lfs merge=lfs -text
+*.xz filter=lfs diff=lfs merge=lfs -text
+*.zip filter=lfs diff=lfs merge=lfs -text
+*.zst filter=lfs diff=lfs merge=lfs -text
+*tfevents* filter=lfs diff=lfs merge=lfs -text
+onnx/decoder_model_merged_q4.onnx_data filter=lfs diff=lfs merge=lfs -text
+onnx/embed_tokens_quantized.onnx_data filter=lfs diff=lfs merge=lfs -text
+onnx/vision_encoder_quantized.onnx_data filter=lfs diff=lfs merge=lfs -text

README.md

@@ -0,0 +1,99 @@
+---
+license: other
+license_name: lfm-open-license-v1.0
+license_link: https://huggingface.co/LiquidAI/LFM2.5-VL-450M/blob/main/LICENSE
+base_model: LiquidAI/LFM2.5-VL-450M
+pipeline_tag: image-text-to-text
+library_name: transformers.js
+tags:
+  - ad-detection
+  - ad-blocking
+  - onnx
+  - webgpu
+  - transformers-js
+  - lfm2.5
+  - vision-language-model
+  - browser
+---
+
+# Minus-v0.1 — a vision model that blocks ads by *looking* at them
+
+**Minus-v0.1** is a 450M-parameter vision–language model fine-tuned to answer one question about an image: **"Is this an advertisement?"** It powers the [Minus Chrome extension](https://github.com/garagehq/Minus-chrome-extension), where it runs **entirely in the browser** (transformers.js + ONNX Runtime Web on WebGPU) and covers detected ads with language flashcards — no filter lists, no servers, no telemetry. It is the browser sibling of the [minus](https://github.com/garagehq/minus) HDMI ad-blocking device.
+
+Instead of matching URLs or DOM patterns like a classic ad blocker, Minus classifies **pixels**: screenshots of page elements, video frames, and iframes. That means it generalizes to first-party ads, sponsored tiles, native placements and streaming-TV commercials that filter lists can't see — and it keeps working when ad-tech rotates domains.
+
+- **Base model:** [LiquidAI/LFM2.5-VL-450M](https://huggingface.co/LiquidAI/LFM2.5-VL-450M)
+- **Task:** binary ad / not-ad classification via a single-token answer (`Yes` / `No`), scored as `p(ad) = P(Yes)`
+- **Prompt:** `Is this an advertisement? Answer Yes or No.`
+- **Format in this repo:** ONNX, quantized for the browser (~430 MB total): q4 decoder, q8 token embeddings, q8 vision encoder. **WebGPU required** (the q4 decoder uses `GatherBlockQuantized`, which onnxruntime-web's WASM backend doesn't implement).
+
+## How it was trained
+
+Minus-v0.1 is **iteration 28** of a months-long training campaign (the first 27 iterations were exploration: base-model selection between FastVLM / SmolVLM / CLIP-style classifiers / LFM2.5-VL, then successive rounds of hard-negative and hard-positive mining, each gated by frozen benchmarks and *live* in-browser soak tests).
+
+**Recipe (frozen across late iterations):** LoRA (r=16, α=32) on the language-model blocks only — vision tower and projector stay frozen — lr 2e-4, 3 epochs, effective batch 32. Trained on a single NVIDIA Jetson AGX Thor (128 GB unified memory). The LoRA is merged before export.
+
+**Data (not released):** ~81,600 training samples across two domains:
+
+- **Streaming TV** — frames sampled by the minus HDMI device during real viewing sessions: commercials vs. program content. The freshest batch (19,328 frames) went through a **two-judge audit** (previous model + an independent VLM judge, disagreements human-reviewed) before training; notably, one device-labeled bucket turned out to be **72% mislabeled** and was corrected — training on raw labels would have poisoned the ad class.
+- **Web pages** — display/banner ad creatives vs. hard negatives mined from real browsing: editorial content, product photography (e-commerce tiles that *look* like ads), site self-promo and UI elements, cookie/consent banners, chat widgets, site headers, and scale-jittered variants of all of these to survive browser resampling.
+
+Roughly **45% of the web-ad positives are native/chum-box style** (Taboola/Outbrain-like), which classic blockers struggle with.
+
+## Evaluation
+
+All gates are held-out sets that were **frozen before** this iteration trained; the live number comes from headed-browser soaks on real sites.
+
+| benchmark | result |
+|---|---|
+| Streaming holdout (1,956 frames, hand-verified) | **99.90%** ad recall / **98.06%** non-ad recall |
+| Static-web bench (999 images) @ shipping gate | **98.0%** ad recall, **11** false positives |
+| Product-image FP holdout (199 ad-look-alike product shots) | **1/199** false positives |
+| Live in-browser precision (month of soak tests, real sites) | **~90–94%** of covered elements are actually ads |
+
+The static-web PR curve dominates or ties the previous production model at **every** operating point with non-ad recall ≥ 95%.
+
+## Using the model
+
+### transformers.js (what the extension does)
+
+```js
+import { AutoProcessor, AutoModelForVision2Seq, RawImage } from "@huggingface/transformers";
+
+const processor = await AutoProcessor.from_pretrained("TheGarageDev/Minus-v0.1");
+const model = await AutoModelForVision2Seq.from_pretrained("TheGarageDev/Minus-v0.1", {
+  device: "webgpu",           // REQUIRED — the q4 decoder is WebGPU-only
+  dtype: { embed_tokens: "q8", vision_encoder: "q8", decoder_model_merged: "q4" },
+});
+
+const image = await RawImage.read(imageUrlOrCanvas);
+const messages = [{ role: "user", content: [
+  { type: "image" },
+  { type: "text", text: "Is this an advertisement? Answer Yes or No." },
+]}];
+const prompt = processor.apply_chat_template(messages, { add_generation_prompt: true });
+const inputs = await processor(prompt, image);
+
+// One decode step; compare the logits of "Yes" vs "No" for a calibrated p(ad).
+const { logits } = await model({ ...inputs });
+// p_ad = softmax over {logit("Yes"), logit("No")} — see the extension's engine
+// (offscreen.js) for the exact token ids + scoring code.
+```
+
+**Thresholding matters.** The extension does not block at p ≥ 0.5 — it uses per-context gates chosen from the PR curve: **0.60** for elements with ad context (iframes / ad-slot containers) and **0.88** for bare images. If you deploy this model, pick your own operating point for your precision target.
+
+### Python (reference / server-side)
+
+The ONNX graphs run under `onnxruntime` too (CPU/CUDA execution providers support the quantized ops). For full-precision experiments, start from the base model and the recipe above.
+
+## Limitations & honest notes
+
+- **Binary classifier, not a chat model.** The fine-tune deliberately collapses the model onto Yes/No answers for one prompt; don't expect general VLM behavior.
+- **Domain:** English-centric web pages and US/EU streaming TV. Ads in other scripts/markets will work worse.
+- **Preprocessing sensitivity:** browser-side image resampling can shift scores on borderline UI-like inputs (we train with scale-jitter to mitigate; a handful of known residuals remain, e.g. certain consent banners).
+- **Screenshots of ads vs. ads:** the model sees pixels. Editorial *about* an ad, or a screenshot of an ad inside an article, can legitimately score high.
+- The training dataset is **not** released.
+
+## License
+
+Inherits the **[LFM Open License v1.0](https://huggingface.co/LiquidAI/LFM2.5-VL-450M/blob/main/LICENSE)** from the base model ([LiquidAI/LFM2.5-VL-450M](https://huggingface.co/LiquidAI/LFM2.5-VL-450M)). The Minus extension source is at [garagehq/Minus-chrome-extension](https://github.com/garagehq/Minus-chrome-extension).

chat_template.jinja

@@ -0,0 +1,92 @@
+{{- bos_token -}}
+{%- set keep_past_thinking = keep_past_thinking | default(false) -%}
+
+{%- macro format_arg_value(arg_value) -%}
+    {%- if arg_value is string -%}
+        {{- '"' + arg_value + '"' -}}
+    {%- elif arg_value is mapping -%}
+        {{- arg_value | tojson -}}
+    {%- else -%}
+        {{- arg_value | string -}}
+    {%- endif -%}
+{%- endmacro -%}
+
+{%- macro parse_content(content) -%}
+    {%- if content is string -%}
+        {{- content -}}
+    {%- else -%}
+        {%- set _ns = namespace(result="") -%}
+        {%- for item in content -%}
+            {%- if item.type == "image" -%}
+                {%- set _ns.result = _ns.result + "<image>" -%}
+            {%- elif item.type == "text" -%}
+                {%- set _ns.result = _ns.result + item.text -%}
+            {%- else -%}
+                {%- set _ns.result = _ns.result + item | tojson -%}
+            {%- endif -%}
+        {%- endfor -%}
+        {{- _ns.result -}}
+    {%- endif -%}
+{%- endmacro -%}
+
+{%- macro render_tool_calls(tool_calls) -%}
+    {%- set tool_calls_ns = namespace(tool_calls=[]) -%}
+    {%- for tool_call in tool_calls -%}
+        {%- set func_name = tool_call.function.name -%}
+        {%- set func_args = tool_call.function.arguments -%}
+        {%- set args_ns = namespace(arg_strings=[]) -%}
+        {%- for arg_name, arg_value in func_args.items() -%}
+            {%- set args_ns.arg_strings = args_ns.arg_strings + [arg_name + "=" + format_arg_value(arg_value)] -%}
+        {%- endfor -%}
+        {%- set tool_calls_ns.tool_calls = tool_calls_ns.tool_calls + [func_name + "(" + (args_ns.arg_strings | join(", ")) + ")"] -%}
+    {%- endfor -%}
+    {{- "<|tool_call_start|>[" + (tool_calls_ns.tool_calls | join(", ")) + "]<|tool_call_end|>" -}}
+{%- endmacro -%}
+
+{%- set ns = namespace(system_prompt="", last_assistant_index=-1) -%}
+{%- if messages[0].role == "system" -%}
+    {%- if messages[0].content is defined -%}
+        {%- set ns.system_prompt = parse_content(messages[0].content) -%}
+    {%- endif -%}
+    {%- set messages = messages[1:] -%}
+{%- endif -%}
+{%- if tools -%}
+    {%- set ns.system_prompt = ns.system_prompt + ("\n\n" if ns.system_prompt else "") + "Today's date: " + strftime_now("%Y-%m-%d") + "\n\nList of tools: " + (tools | tojson) -%}
+{%- endif -%}
+{%- if ns.system_prompt -%}
+    {{- "<|im_start|>system\n" + ns.system_prompt + "<|im_end|>\n" -}}
+{%- endif -%}
+{%- for message in messages -%}
+    {%- if message.role == "assistant" -%}
+        {%- set ns.last_assistant_index = loop.index0 -%}
+    {%- endif -%}
+{%- endfor -%}
+{%- for message in messages -%}
+    {{- "<|im_start|>" + message.role + "\n" -}}
+    {%- if message.role == "assistant" -%}
+        {%- generation -%}
+        {%- if message.thinking is defined and (keep_past_thinking or loop.index0 == ns.last_assistant_index) -%}
+            {{- "<think>" + message.thinking + "</think>" -}}
+        {%- endif -%}
+        {%- if message.tool_calls is defined -%}
+            {{- render_tool_calls(message.tool_calls) -}}
+        {%- endif -%}
+        {%- if message.content is defined -%}
+            {%- set content = parse_content(message.content) -%}
+            {%- if not keep_past_thinking and loop.index0 != ns.last_assistant_index -%}
+                {%- if "</think>" in content -%}
+                    {%- set content = content.split("</think>")[-1] | trim -%}
+                {%- endif -%}
+            {%- endif -%}
+            {{- content + ("" if (continue_final_message and loop.last) else "<|im_end|>\n") -}}
+        {%- endif -%}
+        {%- endgeneration -%}
+    {%- else %}
+        {%- if message.content is defined -%}
+            {{- parse_content(message.content) + "<|im_end|>\n" -}}
+        {%- endif -%}
+    {%- endif %}
+{%- endfor -%}
+{%- if add_generation_prompt -%}
+    {{- "<|im_start|>assistant\n" -}}
+{%- endif -%}

config.json

@@ -0,0 +1,108 @@
+{
+  "architectures": [
+    "Lfm2VlForConditionalGeneration"
+  ],
+  "do_image_splitting": true,
+  "downsample_factor": 2,
+  "dtype": "bfloat16",
+  "encoder_patch_size": 16,
+  "image_token_id": 396,
+  "max_image_tokens": 256,
+  "max_pixels_tolerance": 2.0,
+  "max_tiles": 10,
+  "min_image_tokens": 64,
+  "min_tiles": 2,
+  "model_type": "lfm2_vl",
+  "projector_bias": true,
+  "projector_hidden_act": "gelu",
+  "projector_hidden_size": 2048,
+  "projector_use_layernorm": false,
+  "text_config": {
+    "_name_or_path": "LiquidAI/LFM2-350M",
+    "architectures": [
+      "Lfm2ForCausalLM"
+    ],
+    "block_auto_adjust_ff_dim": true,
+    "block_dim": 1024,
+    "block_ffn_dim_multiplier": 1.0,
+    "block_mlp_init_scale": 1.0,
+    "block_multiple_of": 256,
+    "block_norm_eps": 1e-05,
+    "block_out_init_scale": 1.0,
+    "block_use_swiglu": true,
+    "block_use_xavier_init": true,
+    "bos_token_id": 1,
+    "conv_L_cache": 3,
+    "conv_bias": false,
+    "conv_dim": 1024,
+    "conv_dim_out": 1024,
+    "conv_use_xavier_init": true,
+    "dtype": "bfloat16",
+    "eos_token_id": 7,
+    "full_attn_idxs": null,
+    "hidden_size": 1024,
+    "initializer_range": 0.02,
+    "intermediate_size": 6656,
+    "layer_types": [
+      "conv",
+      "conv",
+      "full_attention",
+      "conv",
+      "conv",
+      "full_attention",
+      "conv",
+      "conv",
+      "full_attention",
+      "conv",
+      "full_attention",
+      "conv",
+      "full_attention",
+      "conv",
+      "full_attention",
+      "conv"
+    ],
+    "max_position_embeddings": 128000,
+    "model_type": "lfm2",
+    "norm_eps": 1e-05,
+    "num_attention_heads": 16,
+    "num_heads": 16,
+    "num_hidden_layers": 16,
+    "num_key_value_heads": 8,
+    "pad_token_id": 0,
+    "rope_parameters": {
+      "rope_theta": 1000000.0,
+      "rope_type": "default"
+    },
+    "tie_word_embeddings": true,
+    "use_cache": true,
+    "use_pos_enc": true,
+    "vocab_size": 65536
+  },
+  "tie_word_embeddings": true,
+  "tile_size": 512,
+  "transformers_version": "5.7.0.dev0",
+  "use_image_special_tokens": true,
+  "use_thumbnail": true,
+  "vision_config": {
+    "attention_dropout": 0.0,
+    "dtype": "bfloat16",
+    "hidden_act": "gelu_pytorch_tanh",
+    "hidden_size": 768,
+    "intermediate_size": 3072,
+    "layer_norm_eps": 1e-06,
+    "model_type": "siglip2_vision_model",
+    "num_attention_heads": 12,
+    "num_channels": 3,
+    "num_hidden_layers": 12,
+    "num_patches": 256,
+    "patch_size": 16,
+    "vision_use_head": false
+  },
+  "transformers.js_config": {
+    "use_external_data_format": true,
+    "kv_cache_dtype": {
+      "fp16": "float16",
+      "q4f16": "float16"
+    }
+  }
+}

generation_config.json

@@ -0,0 +1,7 @@
+{
+  "_from_model_config": true,
+  "bos_token_id": 1,
+  "eos_token_id": 7,
+  "pad_token_id": 0,
+  "transformers_version": "5.7.0.dev0"
+}

onnx/decoder_model_merged_q4.onnx

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:22527f6d3e1af239dac22c8cfd073827f2f7f2b177233b8621c121d1e83ec2e7
+size 188262

onnx/decoder_model_merged_q4.onnx_data

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:791ef1847e833a5c97320de5db55fac7306ad1d49f4b088697e4e169e9d49281
+size 260075520

onnx/embed_tokens_quantized.onnx

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8198d1e54e70e576f9f53efb310fa20fea4674d9e2f55de1d246f194992acd08
+size 822

onnx/embed_tokens_quantized.onnx_data

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:881d06041a181ce14a3d1235a4efe164d639b4f3b579cd2994faf46afa6a5d6b
+size 77594624

onnx/vision_encoder_quantized.onnx

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c8c1991efb42f97689b53c338f393ca21611d6b4d3604ae4476826284a9c0e3e
+size 187977

onnx/vision_encoder_quantized.onnx_data

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8b6b008a1f1743dd08606dd9445b03801af82ee864a7b7e0d47d68237f160987
+size 109874176

preprocessor_config.json

@@ -0,0 +1,37 @@
+{
+  "data_format": "channels_first",
+  "do_image_splitting": true,
+  "do_normalize": true,
+  "do_pad": true,
+  "do_rescale": true,
+  "do_resize": true,
+  "downsample_factor": 2,
+  "encoder_patch_size": 16,
+  "image_mean": [
+    0.5,
+    0.5,
+    0.5
+  ],
+  "image_processor_type": "Lfm2VlImageProcessor",
+  "image_std": [
+    0.5,
+    0.5,
+    0.5
+  ],
+  "max_image_tokens": 256,
+  "max_num_patches": 1024,
+  "max_pixels_tolerance": 2.0,
+  "max_tiles": 10,
+  "min_image_tokens": 64,
+  "min_tiles": 2,
+  "resample": 2,
+  "rescale_factor": 0.00392156862745098,
+  "return_row_col_info": true,
+  "size": {
+    "height": 512,
+    "width": 512
+  },
+  "tile_size": 512,
+  "use_thumbnail": true,
+  "processor_class": "Lfm2VlProcessor"
+}

processor_config.json

@@ -0,0 +1,39 @@
+{
+  "image_processor": {
+    "data_format": "channels_first",
+    "do_image_splitting": true,
+    "do_normalize": true,
+    "do_pad": true,
+    "do_rescale": true,
+    "do_resize": true,
+    "downsample_factor": 2,
+    "encoder_patch_size": 16,
+    "image_mean": [
+      0.5,
+      0.5,
+      0.5
+    ],
+    "image_processor_type": "Lfm2VlImageProcessor",
+    "image_std": [
+      0.5,
+      0.5,
+      0.5
+    ],
+    "max_image_tokens": 256,
+    "max_num_patches": 1024,
+    "max_pixels_tolerance": 2.0,
+    "max_tiles": 10,
+    "min_image_tokens": 64,
+    "min_tiles": 2,
+    "resample": 2,
+    "rescale_factor": 0.00392156862745098,
+    "return_row_col_info": true,
+    "size": {
+      "height": 512,
+      "width": 512
+    },
+    "tile_size": 512,
+    "use_thumbnail": true
+  },
+  "processor_class": "Lfm2VlProcessor"
+}

tokenizer_config.json

@@ -0,0 +1,29 @@
+{
+  "backend": "tokenizers",
+  "bos_token": "<|startoftext|>",
+  "clean_up_tokenization_spaces": true,
+  "eos_token": "<|im_end|>",
+  "extra_special_tokens": [],
+  "image_end_token": "<|image_end|>",
+  "image_start_token": "<|image_start|>",
+  "image_thumbnail": "<|img_thumbnail|>",
+  "image_token": "<image>",
+  "is_local": false,
+  "legacy": false,
+  "local_files_only": false,
+  "model_max_length": 1000000000000000019884624838656,
+  "model_specific_special_tokens": {
+    "image_end_token": "<|image_end|>",
+    "image_start_token": "<|image_start|>",
+    "image_token": "<image>"
+  },
+  "pad_token": "<|pad|>",
+  "processor_class": "Lfm2VlProcessor",
+  "return_token_type_ids": false,
+  "sp_model_kwargs": {},
+  "spaces_between_special_tokens": false,
+  "tokenizer_class": "TokenizersBackend",
+  "use_default_system_prompt": false,
+  "use_fast": true,
+  "chat_template": "{{- bos_token -}}\n{%- set keep_past_thinking = keep_past_thinking | default(false) -%}\n\n{%- macro format_arg_value(arg_value) -%}\n    {%- if arg_value is string -%}\n        {{- '\"' + arg_value + '\"' -}}\n    {%- elif arg_value is mapping -%}\n        {{- arg_value | tojson -}}\n    {%- else -%}\n        {{- arg_value | string -}}\n    {%- endif -%}\n{%- endmacro -%}\n\n{%- macro parse_content(content) -%}\n    {%- if content is string -%}\n        {{- content -}}\n    {%- else -%}\n        {%- set _ns = namespace(result=\"\") -%}\n        {%- for item in content -%}\n            {%- if item.type == \"image\" -%}\n                {%- set _ns.result = _ns.result + \"<image>\" -%}\n            {%- elif item.type == \"text\" -%}\n                {%- set _ns.result = _ns.result + item.text -%}\n            {%- else -%}\n                {%- set _ns.result = _ns.result + item | tojson -%}\n            {%- endif -%}\n        {%- endfor -%}\n        {{- _ns.result -}}\n    {%- endif -%}\n{%- endmacro -%}\n\n{%- macro render_tool_calls(tool_calls) -%}\n    {%- set tool_calls_ns = namespace(tool_calls=[]) -%}\n    {%- for tool_call in tool_calls -%}\n        {%- set func_name = tool_call.function.name -%}\n        {%- set func_args = tool_call.function.arguments -%}\n        {%- set args_ns = namespace(arg_strings=[]) -%}\n        {%- for arg_name, arg_value in func_args.items() -%}\n            {%- set args_ns.arg_strings = args_ns.arg_strings + [arg_name + \"=\" + format_arg_value(arg_value)] -%}\n        {%- endfor -%}\n        {%- set tool_calls_ns.tool_calls = tool_calls_ns.tool_calls + [func_name + \"(\" + (args_ns.arg_strings | join(\", \")) + \")\"] -%}\n    {%- endfor -%}\n    {{- \"<|tool_call_start|>[\" + (tool_calls_ns.tool_calls | join(\", \")) + \"]<|tool_call_end|>\" -}}\n{%- endmacro -%}\n\n{%- set ns = namespace(system_prompt=\"\", last_assistant_index=-1) -%}\n{%- if messages[0].role == \"system\" -%}\n    {%- if messages[0].content is defined -%}\n        {%- set ns.system_prompt = parse_content(messages[0].content) -%}\n    {%- endif -%}\n    {%- set messages = messages[1:] -%}\n{%- endif -%}\n{%- if tools -%}\n    {%- set ns.system_prompt = ns.system_prompt + (\"\\n\\n\" if ns.system_prompt else \"\") + \"Today's date: \" + strftime_now(\"%Y-%m-%d\") + \"\\n\\nList of tools: \" + (tools | tojson) -%}\n{%- endif -%}\n{%- if ns.system_prompt -%}\n    {{- \"<|im_start|>system\\n\" + ns.system_prompt + \"<|im_end|>\\n\" -}}\n{%- endif -%}\n{%- for message in messages -%}\n    {%- if message.role == \"assistant\" -%}\n        {%- set ns.last_assistant_index = loop.index0 -%}\n    {%- endif -%}\n{%- endfor -%}\n{%- for message in messages -%}\n    {{- \"<|im_start|>\" + message.role + \"\\n\" -}}\n    {%- if message.role == \"assistant\" -%}\n        {%- generation -%}\n        {%- if message.thinking is defined and (keep_past_thinking or loop.index0 == ns.last_assistant_index) -%}\n            {{- \"<think>\" + message.thinking + \"</think>\" -}}\n        {%- endif -%}\n        {%- if message.tool_calls is defined -%}\n            {{- render_tool_calls(message.tool_calls) -}}\n        {%- endif -%}\n        {%- if message.content is defined -%}\n            {%- set content = parse_content(message.content) -%}\n            {%- if not keep_past_thinking and loop.index0 != ns.last_assistant_index -%}\n                {%- if \"</think>\" in content -%}\n                    {%- set content = content.split(\"</think>\")[-1] | trim -%}\n                {%- endif -%}\n            {%- endif -%}\n            {{- content + (\"\" if (continue_final_message and loop.last) else \"<|im_end|>\\n\") -}}\n        {%- endif -%}\n        {%- endgeneration -%}\n    {%- else %}\n        {%- if message.content is defined -%}\n            {{- parse_content(message.content) + \"<|im_end|>\\n\" -}}\n        {%- endif -%}\n    {%- endif %}\n{%- endfor -%}\n{%- if add_generation_prompt -%}\n    {{- \"<|im_start|>assistant\\n\" -}}\n{%- endif -%}"
+}