Sync gemma-3-1b-it-int8-v1

gemma-3-1b-it-int8-v1/README.md

@@ -0,0 +1,48 @@
+---
+library_name: cellm
+tags:
+- mobile
+- rust
+- memory-efficient
+- quantized
+- gemma
+---
+
+# Gemma 3 1B IT (Cellm Int8)
+
+This folder contains a Cellm-converted Gemma 3 1B Instruct model and tokenizer assets, ready for publishing to Hugging Face.
+
+## Files
+- `gemma-3-1b-it-int8-v1.cellm`
+- `tokenizer.json`
+- `tokenizer_config.json`
+- `chat_template.jinja`
+
+## Model Details
+- **Base model**: `google/gemma-3-1b-it`
+- **Format**: `.cellm`
+- **Quantization**: INT8 symmetric weight-only
+- **Size**: ~1.2 GB
+
+## Inference Check (cellm)
+
+```bash
+cd /Users/jeff/Desktop/cellm
+./target/release/infer \
+  --model models/to-huggingface/gemma-3-1b-it-int8-v1/gemma-3-1b-it-int8-v1.cellm \
+  --tokenizer models/to-huggingface/gemma-3-1b-it-int8-v1/tokenizer.json \
+  --prompt "what's twitch.com?" \
+  --chat \
+  --chat-format plain \
+  --gen 48 \
+  --temperature 0 \
+  --backend cpu \
+  --kv-encoding f16
+```
+
+## Notes
+- This INT8 variant produced coherent output in local validation.
+- INT4 variant was smaller (~481 MB) but quality was significantly worse.
+
+## License
+Subject to Gemma terms and upstream license constraints.

gemma-3-1b-it-int8-v1/chat_template.jinja

@@ -0,0 +1,47 @@
+{{ bos_token }}
+{%- if messages[0]['role'] == 'system' -%}
+    {%- if messages[0]['content'] is string -%}
+        {%- set first_user_prefix = messages[0]['content'] + '
+
+' -%}
+    {%- else -%}
+        {%- set first_user_prefix = messages[0]['content'][0]['text'] + '
+
+' -%}
+    {%- endif -%}
+    {%- set loop_messages = messages[1:] -%}
+{%- else -%}
+    {%- set first_user_prefix = "" -%}
+    {%- set loop_messages = messages -%}
+{%- endif -%}
+{%- for message in loop_messages -%}
+    {%- if (message['role'] == 'user') != (loop.index0 % 2 == 0) -%}
+        {{ raise_exception("Conversation roles must alternate user/assistant/user/assistant/...") }}
+    {%- endif -%}
+    {%- if (message['role'] == 'assistant') -%}
+        {%- set role = "model" -%}
+    {%- else -%}
+        {%- set role = message['role'] -%}
+    {%- endif -%}
+    {{ '<start_of_turn>' + role + '
+' + (first_user_prefix if loop.first else "") }}
+    {%- if message['content'] is string -%}
+        {{ message['content'] | trim }}
+    {%- elif message['content'] is iterable -%}
+        {%- for item in message['content'] -%}
+            {%- if item['type'] == 'image' -%}
+                {{ '<start_of_image>' }}
+            {%- elif item['type'] == 'text' -%}
+                {{ item['text'] | trim }}
+            {%- endif -%}
+        {%- endfor -%}
+    {%- else -%}
+        {{ raise_exception("Invalid content type") }}
+    {%- endif -%}
+    {{ '<end_of_turn>
+' }}
+{%- endfor -%}
+{%- if add_generation_prompt -%}
+    {{'<start_of_turn>model
+'}}
+{%- endif -%}

gemma-3-1b-it-int8-v1/gemma-3-1b-it-int8-v1.cellm

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:73cc1fe14164fc1ac208c7c86ead0f1a56d07375f0da52461c282f954ec623d8
+size 1302993856

gemma-3-1b-it-int8-v1/tokenizer.json

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4667f2089529e8e7657cfb6d1c19910ae71ff5f28aa7ab2ff2763330affad795
+size 33384568