Upload model

.gitattributes

@@ -33,3 +33,13 @@ saved_model/**/* 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
+gemma3n-e4b-it-afq3-0.uqff filter=lfs diff=lfs merge=lfs -text
+gemma3n-e4b-it-afq4-0.uqff filter=lfs diff=lfs merge=lfs -text
+gemma3n-e4b-it-afq6-0.uqff filter=lfs diff=lfs merge=lfs -text
+gemma3n-e4b-it-afq8-0.uqff filter=lfs diff=lfs merge=lfs -text
+gemma3n-e4b-it-f8e4m3-0.uqff filter=lfs diff=lfs merge=lfs -text
+gemma3n-e4b-it-q3k-0.uqff filter=lfs diff=lfs merge=lfs -text
+gemma3n-e4b-it-q4k-0.uqff filter=lfs diff=lfs merge=lfs -text
+gemma3n-e4b-it-q5k-0.uqff filter=lfs diff=lfs merge=lfs -text
+gemma3n-e4b-it-q8_0-0.uqff filter=lfs diff=lfs merge=lfs -text
+tokenizer.json filter=lfs diff=lfs merge=lfs -text

README.md

@@ -0,0 +1,32 @@
+---
+tags:
+  - uqff
+  - mistral.rs
+base_model: google/gemma-3n-E4B-it
+base_model_relation: quantized
+---
+
+<!-- Autogenerated from user input. -->
+
+# `google/gemma-3n-E4B-it`, UQFF quantization
+
+
+Run with [mistral.rs](https://github.com/EricLBuehler/mistral.rs). Documentation: [UQFF docs](https://github.com/EricLBuehler/mistral.rs/blob/master/docs/UQFF.md).
+
+1) **Flexible** 🌀: Multiple quantization formats in *one* file format with *one* framework to run them all.
+2) **Reliable** 🔒: Compatibility ensured with *embedded* and *checked* semantic versioning information from day 1.
+3) **Easy** 🤗: Download UQFF models *easily* and *quickly* from Hugging Face, or use a local file.
+3) **Customizable** 🛠️: Make and publish your own UQFF files in minutes.
+
+## Examples
+|Quantization type(s)|Example|
+|--|--|
+|AFQ3|`./mistralrs-server -i vision-plain -m EricB/gemma-3n-E4B-it-UQFF -f gemma3n-e4b-it-afq3-0.uqff`|
+|AFQ4|`./mistralrs-server -i vision-plain -m EricB/gemma-3n-E4B-it-UQFF -f gemma3n-e4b-it-afq4-0.uqff`|
+|AFQ6|`./mistralrs-server -i vision-plain -m EricB/gemma-3n-E4B-it-UQFF -f gemma3n-e4b-it-afq6-0.uqff`|
+|AFQ8|`./mistralrs-server -i vision-plain -m EricB/gemma-3n-E4B-it-UQFF -f gemma3n-e4b-it-afq8-0.uqff`|
+|F8E4M3|`./mistralrs-server -i vision-plain -m EricB/gemma-3n-E4B-it-UQFF -f gemma3n-e4b-it-f8e4m3-0.uqff`|
+|Q3K|`./mistralrs-server -i vision-plain -m EricB/gemma-3n-E4B-it-UQFF -f gemma3n-e4b-it-q3k-0.uqff`|
+|Q4K|`./mistralrs-server -i vision-plain -m EricB/gemma-3n-E4B-it-UQFF -f gemma3n-e4b-it-q4k-0.uqff`|
+|Q5K|`./mistralrs-server -i vision-plain -m EricB/gemma-3n-E4B-it-UQFF -f gemma3n-e4b-it-q5k-0.uqff`|
+|Q8_0|`./mistralrs-server -i vision-plain -m EricB/gemma-3n-E4B-it-UQFF -f gemma3n-e4b-it-q8_0-0.uqff`|

chat_template.jinja

@@ -0,0 +1,49 @@
+{{ 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'] == 'audio' -%}
+                {{ '<audio_soft_token>' }}
+            {%- elif item['type'] == 'image' -%}
+                {{ '<image_soft_token>' }}
+            {%- 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 -%}

config.json

@@ -0,0 +1,189 @@
+{
+  "architectures": [
+    "Gemma3nForConditionalGeneration"
+  ],
+  "audio_config": {
+    "conf_attention_chunk_size": 12,
+    "conf_attention_context_left": 13,
+    "conf_attention_context_right": 0,
+    "conf_attention_logit_cap": 50.0,
+    "conf_conv_kernel_size": 5,
+    "conf_num_attention_heads": 8,
+    "conf_num_hidden_layers": 12,
+    "conf_positional_bias_size": 256,
+    "conf_reduction_factor": 4,
+    "conf_residual_weight": 0.5,
+    "gradient_clipping": 10000000000.0,
+    "hidden_size": 1536,
+    "input_feat_size": 128,
+    "model_type": "gemma3n_audio",
+    "rms_norm_eps": 1e-06,
+    "sscp_conv_channel_size": [
+      128,
+      32
+    ],
+    "sscp_conv_eps": 0.001,
+    "sscp_conv_kernel_size": [
+      [
+        3,
+        3
+      ],
+      [
+        3,
+        3
+      ]
+    ],
+    "sscp_conv_stride_size": [
+      [
+        2,
+        2
+      ],
+      [
+        2,
+        2
+      ]
+    ],
+    "torch_dtype": "bfloat16",
+    "vocab_offset": 262272,
+    "vocab_size": 128
+  },
+  "audio_soft_tokens_per_image": 188,
+  "audio_token_id": 262273,
+  "boa_token_id": 256000,
+  "boi_token_id": 255999,
+  "eoa_token_id": 262272,
+  "eoi_token_id": 262144,
+  "eos_token_id": [
+    1,
+    106
+  ],
+  "image_token_id": 262145,
+  "initializer_range": 0.02,
+  "model_type": "gemma3n",
+  "text_config": {
+    "activation_sparsity_pattern": [
+      0.95,
+      0.95,
+      0.95,
+      0.95,
+      0.95,
+      0.95,
+      0.95,
+      0.95,
+      0.95,
+      0.95,
+      0.0,
+      0.0,
+      0.0,
+      0.0,
+      0.0,
+      0.0,
+      0.0,
+      0.0,
+      0.0,
+      0.0,
+      0.0,
+      0.0,
+      0.0,
+      0.0,
+      0.0,
+      0.0,
+      0.0,
+      0.0,
+      0.0,
+      0.0,
+      0.0,
+      0.0,
+      0.0,
+      0.0,
+      0.0
+    ],
+    "altup_active_idx": 0,
+    "altup_coef_clip": 120.0,
+    "altup_correct_scale": true,
+    "altup_lr_multiplier": 1.0,
+    "altup_num_inputs": 4,
+    "attention_bias": false,
+    "attention_dropout": 0.0,
+    "final_logit_softcapping": 30.0,
+    "head_dim": 256,
+    "hidden_activation": "gelu_pytorch_tanh",
+    "hidden_size": 2048,
+    "hidden_size_per_layer_input": 256,
+    "initializer_range": 0.02,
+    "intermediate_size": 16384,
+    "laurel_rank": 64,
+    "layer_types": [
+      "sliding_attention",
+      "sliding_attention",
+      "sliding_attention",
+      "sliding_attention",
+      "full_attention",
+      "sliding_attention",
+      "sliding_attention",
+      "sliding_attention",
+      "sliding_attention",
+      "full_attention",
+      "sliding_attention",
+      "sliding_attention",
+      "sliding_attention",
+      "sliding_attention",
+      "full_attention",
+      "sliding_attention",
+      "sliding_attention",
+      "sliding_attention",
+      "sliding_attention",
+      "full_attention",
+      "sliding_attention",
+      "sliding_attention",
+      "sliding_attention",
+      "sliding_attention",
+      "full_attention",
+      "sliding_attention",
+      "sliding_attention",
+      "sliding_attention",
+      "sliding_attention",
+      "full_attention",
+      "sliding_attention",
+      "sliding_attention",
+      "sliding_attention",
+      "sliding_attention",
+      "full_attention"
+    ],
+    "max_position_embeddings": 32768,
+    "model_type": "gemma3n_text",
+    "num_attention_heads": 8,
+    "num_hidden_layers": 35,
+    "num_key_value_heads": 2,
+    "num_kv_shared_layers": 15,
+    "query_pre_attn_scalar": 256,
+    "rms_norm_eps": 1e-06,
+    "rope_local_base_freq": 10000.0,
+    "rope_scaling": null,
+    "rope_theta": 1000000.0,
+    "sliding_window": 512,
+    "torch_dtype": "bfloat16",
+    "use_cache": true,
+    "vocab_size": 262400,
+    "vocab_size_per_layer_input": 262144
+  },
+  "torch_dtype": "bfloat16",
+  "transformers_version": "4.53.0.dev0",
+  "vision_config": {
+    "architecture": "mobilenetv5_300m_enc",
+    "do_pooling": true,
+    "hidden_size": 2048,
+    "initializer_range": 0.02,
+    "label_names": [
+      "LABEL_0",
+      "LABEL_1"
+    ],
+    "model_type": "gemma3n_vision",
+    "num_classes": 2,
+    "rms_norm_eps": 1e-06,
+    "torch_dtype": "bfloat16",
+    "vocab_offset": 262144,
+    "vocab_size": 128
+  },
+  "vision_soft_tokens_per_image": 256
+}

gemma3n-e4b-it-afq3-0.uqff

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e59177bd53812542dbd594a1a6e5512a2bbf99d9a5ce5e070864f53d6604efc0
+size 2257345568

gemma3n-e4b-it-afq4-0.uqff

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:df6fd197ea7ecdc53529b2af00b75e3e4504824e7c6427c265ec61fa33682591
+size 2902285408

gemma3n-e4b-it-afq6-0.uqff

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:657181fda962af64abea7a2de917e92a44080d63387e6d831f14f6e65c788b69
+size 4192165032

gemma3n-e4b-it-afq8-0.uqff

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d0d4a6da96e3484bb64603a8b5ca596d81b36d77954876e0e614406ce1042925
+size 5482044736

gemma3n-e4b-it-f8e4m3-0.uqff

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d99f19e17ab63f8e329f47d2d3a04ec38c4888f2ff78cbc0c335790222dc29e5
+size 5159564704

gemma3n-e4b-it-q3k-0.uqff

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3889a2ddef8611c4371b8cc5db6bd7efac1086428399122813a6939403c1347a
+size 2217020528

gemma3n-e4b-it-q4k-0.uqff

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ad7d848f1a304bad92516ce6e047a113583085644a2c124a54556b6c2297b452
+size 2902269112

gemma3n-e4b-it-q5k-0.uqff

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:96c29f8c1f076d744e23e554f307c05acf38652aff4a73d5dc7212db6e2687c2
+size 3547208928

gemma3n-e4b-it-q8_0-0.uqff

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a0cb4a94faa44bae7137a8f0544b07d0b90f6a9346ebd2bacef1fe53f6e92d04
+size 5482028440

generation_config.json

@@ -0,0 +1,13 @@
+{
+  "bos_token_id": 2,
+  "cache_implementation": "hybrid",
+  "do_sample": true,
+  "eos_token_id": [
+    1,
+    106
+  ],
+  "pad_token_id": 0,
+  "top_k": 64,
+  "top_p": 0.95,
+  "transformers_version": "4.53.0.dev0"
+}

preprocessor_config.json

@@ -0,0 +1,51 @@
+{
+  "crop_size": null,
+  "data_format": "channels_first",
+  "default_to_square": false,
+  "device": null,
+  "dither": 0.0,
+  "do_center_crop": null,
+  "do_convert_rgb": null,
+  "do_normalize": false,
+  "do_rescale": true,
+  "do_resize": true,
+  "feature_extractor_type": "Gemma3nAudioFeatureExtractor",
+  "feature_size": 128,
+  "fft_length": 1024,
+  "fft_overdrive": true,
+  "frame_length": 512,
+  "hop_length": 160,
+  "image_mean": [
+    0.5,
+    0.5,
+    0.5
+  ],
+  "image_processor_type": "SiglipImageProcessorFast",
+  "image_seq_length": 256,
+  "image_std": [
+    0.5,
+    0.5,
+    0.5
+  ],
+  "input_data_format": null,
+  "input_scale_factor": 1.0,
+  "max_frequency": 7600.0,
+  "mel_floor": 1e-05,
+  "min_frequency": 125.0,
+  "padding_side": "right",
+  "padding_value": 0.0,
+  "per_bin_mean": null,
+  "per_bin_stddev": null,
+  "preemphasis": 0.97,
+  "preemphasis_htk_flavor": true,
+  "processor_class": "Gemma3nProcessor",
+  "resample": 2,
+  "rescale_factor": 0.00392156862745098,
+  "return_attention_mask": false,
+  "return_tensors": null,
+  "sampling_rate": 16000,
+  "size": {
+    "height": 768,
+    "width": 768
+  }
+}

processor_config.json

@@ -0,0 +1,5 @@
+{
+  "audio_seq_length": 188,
+  "image_seq_length": 256,
+  "processor_class": "Gemma3nProcessor"
+}

residual.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2d9f97b2d005f63ea1115773d528e8889b94a28d7e0b773f946637018fc7e99d
+size 6456156792

tokenizer.json

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b6c35ee648c07754b44cd9e371c75d4caa05c4504910b7ad29b1847ee9d8ba5d
+size 33442553