Instructions to use google/gemma-4-E4B-it with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Transformers
How to use google/gemma-4-E4B-it with Transformers:
# pip install -U transformers accelerate # Load model directly from transformers import AutoProcessor, AutoModelForMultimodalLM processor = AutoProcessor.from_pretrained("google/gemma-4-E4B-it") model = AutoModelForMultimodalLM.from_pretrained("google/gemma-4-E4B-it", device_map="auto") - Notebooks
- Google Colab
- Kaggle
- AMD Developer Cloud
Add ParseBench 2 evaluation results
Browse filesAdd all six ParseBench 2 dimensions and the overall score from public CI replay.
Matching ParseBench PR: https://github.com/run-llama/ParseBench/pull/283
Leaderboard: https://huggingface.co/datasets/llamaindex/ParseBench-2
Format: https://huggingface.co/docs/hub/eval-results
.eval_results/parsebench-2.yaml
ADDED
|
@@ -0,0 +1,70 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
- dataset:
|
| 2 |
+
id: llamaindex/ParseBench-2
|
| 3 |
+
task_id: overall
|
| 4 |
+
value: 42.32
|
| 5 |
+
date: '2026-11-15'
|
| 6 |
+
source:
|
| 7 |
+
url: https://huggingface.co/datasets/llamaindex/ParseBench-2
|
| 8 |
+
name: ParseBench-2
|
| 9 |
+
user: boyang-runllama
|
| 10 |
+
notes: "Pipeline name: gemma4_e4b_vllm_with_layout. Scores from public CI replay."
|
| 11 |
+
- dataset:
|
| 12 |
+
id: llamaindex/ParseBench-2
|
| 13 |
+
task_id: text
|
| 14 |
+
value: 65.07
|
| 15 |
+
date: '2026-11-15'
|
| 16 |
+
source:
|
| 17 |
+
url: https://huggingface.co/datasets/llamaindex/ParseBench-2
|
| 18 |
+
name: ParseBench-2
|
| 19 |
+
user: boyang-runllama
|
| 20 |
+
notes: "Pipeline name: gemma4_e4b_vllm_with_layout. Scores from public CI replay."
|
| 21 |
+
- dataset:
|
| 22 |
+
id: llamaindex/ParseBench-2
|
| 23 |
+
task_id: formatting
|
| 24 |
+
value: 70.57
|
| 25 |
+
date: '2026-11-15'
|
| 26 |
+
source:
|
| 27 |
+
url: https://huggingface.co/datasets/llamaindex/ParseBench-2
|
| 28 |
+
name: ParseBench-2
|
| 29 |
+
user: boyang-runllama
|
| 30 |
+
notes: "Pipeline name: gemma4_e4b_vllm_with_layout. Scores from public CI replay."
|
| 31 |
+
- dataset:
|
| 32 |
+
id: llamaindex/ParseBench-2
|
| 33 |
+
task_id: tables
|
| 34 |
+
value: 30.82
|
| 35 |
+
date: '2026-11-15'
|
| 36 |
+
source:
|
| 37 |
+
url: https://huggingface.co/datasets/llamaindex/ParseBench-2
|
| 38 |
+
name: ParseBench-2
|
| 39 |
+
user: boyang-runllama
|
| 40 |
+
notes: "Pipeline name: gemma4_e4b_vllm_with_layout. Scores from public CI replay."
|
| 41 |
+
- dataset:
|
| 42 |
+
id: llamaindex/ParseBench-2
|
| 43 |
+
task_id: charts
|
| 44 |
+
value: 5.39
|
| 45 |
+
date: '2026-11-15'
|
| 46 |
+
source:
|
| 47 |
+
url: https://huggingface.co/datasets/llamaindex/ParseBench-2
|
| 48 |
+
name: ParseBench-2
|
| 49 |
+
user: boyang-runllama
|
| 50 |
+
notes: "Pipeline name: gemma4_e4b_vllm_with_layout. Scores from public CI replay."
|
| 51 |
+
- dataset:
|
| 52 |
+
id: llamaindex/ParseBench-2
|
| 53 |
+
task_id: layout
|
| 54 |
+
value: 21.57
|
| 55 |
+
date: '2026-11-15'
|
| 56 |
+
source:
|
| 57 |
+
url: https://huggingface.co/datasets/llamaindex/ParseBench-2
|
| 58 |
+
name: ParseBench-2
|
| 59 |
+
user: boyang-runllama
|
| 60 |
+
notes: "Pipeline name: gemma4_e4b_vllm_with_layout. Scores from public CI replay."
|
| 61 |
+
- dataset:
|
| 62 |
+
id: llamaindex/ParseBench-2
|
| 63 |
+
task_id: forms
|
| 64 |
+
value: 60.48
|
| 65 |
+
date: '2026-11-15'
|
| 66 |
+
source:
|
| 67 |
+
url: https://huggingface.co/datasets/llamaindex/ParseBench-2
|
| 68 |
+
name: ParseBench-2
|
| 69 |
+
user: boyang-runllama
|
| 70 |
+
notes: "Pipeline name: gemma4_e4b_vllm_with_layout. Scores from public CI replay."
|