shithumb commited on
Commit
c1106d7
ยท
verified ยท
1 Parent(s): 77b9228

Upload 5 files

Browse files
Files changed (5) hide show
  1. LICENSE +40 -0
  2. Readme.md +76 -0
  3. config.json +36 -0
  4. tokenizer.json +0 -0
  5. venus-x1.bin +3 -0
LICENSE ADDED
@@ -0,0 +1,40 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ Venus-X1 - A Fine-Tuned Falcon-1B Derivative
2
+ =============================================
3
+
4
+ Base Model:
5
+ -----------
6
+ This model, Venus-X1, is a derivative work of the Falcon-RW-1B model released by TII (Technology Innovation Institute), UAE.
7
+
8
+ Original model repository:
9
+ https://huggingface.co/tiiuae/falcon-rw-1b
10
+
11
+ Original model weights were downloaded using:
12
+ wget https://huggingface.co/tiiuae/falcon-rw-1b/resolve/main/pytorch_model.bin
13
+
14
+ License for Base Model:
15
+ -----------------------
16
+ The original Falcon-RW-1B model is licensed under the Apache License 2.0.
17
+ You can find the original license here:
18
+ https://huggingface.co/tiiuae/falcon-rw-1b/blob/main/LICENSE
19
+
20
+ Modifications:
21
+ --------------
22
+ This version (Venus-X1) has been fine-tuned on custom datasets by Shi Thumb (https://huggingface.co/shithumb).
23
+ The fine-tuning includes additional training on domain-specific corpora related to self-development and creative ideation.
24
+
25
+ License for Venus-X1:
26
+ ---------------------
27
+ Venus-X1 is released under the same license as the base model: Apache License 2.0.
28
+ You are free to use, modify, and redistribute this model under the terms of that license.
29
+
30
+ Attribution:
31
+ ------------
32
+ Please include proper attribution to both TII (original Falcon creators) and Shi Thumb when using this model in academic or commercial applications.
33
+
34
+ Contact:
35
+ --------
36
+ For questions or collaborations:
37
+ https://huggingface.co/shithumb
38
+
39
+ Last Updated: August 2025
40
+
Readme.md ADDED
@@ -0,0 +1,76 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # Venus-X1 ๐ŸŒŒ
2
+ A fine-tuned AI model based on [Falcon-RW-1B](https://huggingface.co/tiiuae/falcon-rw-1b), adapted for self-development and creative ideation tasks.
3
+
4
+ ## ๐Ÿ” Overview
5
+
6
+ **Venus-X1** is a lightweight transformer model derived from **Falcon-RW-1B** developed by the Technology Innovation Institute (TII). This model has been fine-tuned using curated datasets on topics such as:
7
+
8
+ - Self-development and introspection
9
+ - Cognitive enhancement
10
+ - Creativity and mindset exploration
11
+ - Prompting frameworks for knowledge work
12
+
13
+ The goal of Venus-X1 is to provide a compact yet insightful model for researchers, creators, and developers working on personal growth and AI-generated self-reflection tools.
14
+
15
+ ---
16
+
17
+ ## ๐Ÿง  Base Model
18
+
19
+ - **Original Name:** Falcon-RW-1B
20
+ - **Creator:** [tiiuae](https://huggingface.co/tiiuae)
21
+ - **License:** Apache 2.0
22
+ - **URL:** [https://huggingface.co/tiiuae/falcon-rw-1b](https://huggingface.co/tiiuae/falcon-rw-1b)
23
+
24
+ Model weights downloaded using:
25
+
26
+ ```bash
27
+ wget https://huggingface.co/tiiuae/falcon-rw-1b/resolve/main/pytorch_model.bin
28
+ ๐Ÿ› ๏ธ Fine-Tuning Details
29
+ Fine-tuned by: Shi Thumb (https://huggingface.co/shithumb)
30
+
31
+ Dataset used: AI_Self_Development_Guide
32
+
33
+ Token length: 2048
34
+
35
+ Epochs: 3โ€“5
36
+
37
+ Tokenizer: Inherits from Falcon-RW-1B tokenizer
38
+
39
+ The model was trained using Hugging Face Transformers with mixed-precision (fp16) on a single GPU.
40
+
41
+ ๐Ÿ’พ How to Use
42
+ python
43
+ Salin
44
+ Edit
45
+ from transformers import AutoModelForCausalLM, AutoTokenizer
46
+
47
+ model = AutoModelForCausalLM.from_pretrained("shithumb/venus-x1")
48
+ tokenizer = AutoTokenizer.from_pretrained("shithumb/venus-x1")
49
+
50
+ prompt = "How can I improve my daily focus and motivation?"
51
+ inputs = tokenizer(prompt, return_tensors="pt")
52
+ outputs = model.generate(**inputs, max_new_tokens=200)
53
+ print(tokenizer.decode(outputs[0], skip_special_tokens=True))
54
+ ๐Ÿ“š Dataset
55
+ Dataset used: venus-1.csv
56
+
57
+ This dataset contains guided self-development prompts and insights in structured format.
58
+
59
+ ๐Ÿ“„ License
60
+ This model is released under the Apache 2.0 license, same as its base model.
61
+
62
+ See LICENSE.txt for details.
63
+
64
+ ๐Ÿค Acknowledgements
65
+ Thanks to tiiuae for open-sourcing Falcon-RW-1B.
66
+
67
+ Fine-tuned and republished by Shi Thumb
68
+
69
+ ๐Ÿงญ Contact
70
+ For questions, discussions, or collaborations, reach out via:
71
+
72
+ Hugging Face: https://huggingface.co/shithumb
73
+
74
+ Email: (add if applicable)
75
+
76
+ Last updated: August 2025
config.json ADDED
@@ -0,0 +1,36 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "alibi": true,
3
+ "apply_residual_connection_post_layernorm": false,
4
+ "architectures": [
5
+ "VenusForCausalLM"
6
+ ],
7
+ "attention_dropout": 0.0,
8
+ "auto_map": {
9
+ "AutoConfig": "configuration_venus.VenusConfig",
10
+ "AutoModel": "modeling_venus.VenusModel",
11
+ "AutoModelForSequenceClassification": "modeling_venus.VenusForSequenceClassification",
12
+ "AutoModelForTokenClassification": "modeling_venus.VenusForTokenClassification",
13
+ "AutoModelForQuestionAnswering": "modeling_venus.VenusForQuestionAnswering",
14
+ "AutoModelForCausalLM": "modeling_venus.VenusForCausalLM"
15
+ },
16
+ "bias": true,
17
+ "bos_token_id": 1,
18
+ "eos_token_id": 2,
19
+ "hidden_dropout": 0.0,
20
+ "hidden_size": 2048,
21
+ "initializer_range": 0.02,
22
+ "layer_norm_epsilon": 1e-05,
23
+ "model_type": "venus",
24
+ "multi_query": false,
25
+ "new_decoder_architecture": false,
26
+ "num_attention_heads": 32,
27
+ "num_hidden_layers": 24,
28
+ "parallel_attn": false,
29
+ "torch_dtype": "bfloat16",
30
+ "transformers_version": "4.27.4",
31
+ "use_cache": true,
32
+ "vocab_size": 50304,
33
+ "model_origin": "Adapted from Falcon-7B architecture (Technology by TII, Apache 2.0 License)",
34
+ "model_name": "Venus-X1"
35
+ }
36
+
tokenizer.json ADDED
The diff for this file is too large to render. See raw diff
 
venus-x1.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:3a0d68f0309c8f7ec913f51edf8bf2eca849477c9f53db727f49ffa2f6019251
3
+ size 2623348889