Upload TinyBrain-100M Instruct model

README.md

@@ -1,3 +1,41 @@
 ---
 license: apache-2.0
+language:
+- en
+base_model: exnivo/tinybrain-100m-base
+datasets:
+- exnivo/tinybrain-instruct
+tags:
+- text-generation
+- causal-lm
+- instruct
+- chat
+- tinybrain
+- sft
 ---
+
+# TinyBrain-100M Instruct
+
+TinyBrain-100M Instruct is a small chat/instruct model fine-tuned from `exnivo/tinybrain-100m-base` on `exnivo/tinybrain-instruct`.
+
+## Model details
+
+- Parameters: ~103M
+- Architecture: LLaMA-style causal transformer
+- Tokenizer: custom TinyBrain byte-level BPE
+- Vocab size: 24,000
+- Base model: `exnivo/tinybrain-100m-base`
+- SFT dataset: `exnivo/tinybrain-instruct`
+
+## Prompt format
+
+```text
+<|bos|><|user|>
+Your message here
+<|end|>
+<|assistant|>
+````
+
+## Limitations
+
+This is a very small model. It can hallucinate, make factual mistakes, fail at math, and give unreliable answers. Do not use it for medical, legal, financial, or safety-critical decisions.

config.json

@@ -0,0 +1,32 @@
+{
+  "architectures": [
+    "LlamaForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "bos_token_id": 1,
+  "dtype": "bfloat16",
+  "eos_token_id": 2,
+  "head_dim": 64,
+  "hidden_act": "silu",
+  "hidden_size": 768,
+  "initializer_range": 0.02,
+  "intermediate_size": 2048,
+  "max_position_embeddings": 2048,
+  "mlp_bias": false,
+  "model_type": "llama",
+  "num_attention_heads": 12,
+  "num_hidden_layers": 12,
+  "num_key_value_heads": 12,
+  "pad_token_id": 0,
+  "pretraining_tp": 1,
+  "rms_norm_eps": 1e-05,
+  "rope_parameters": {
+    "rope_theta": 10000.0,
+    "rope_type": "default"
+  },
+  "tie_word_embeddings": true,
+  "transformers_version": "5.12.1",
+  "use_cache": false,
+  "vocab_size": 24000
+}

generation_config.json

@@ -0,0 +1,10 @@
+{
+  "_from_model_config": true,
+  "bos_token_id": 1,
+  "eos_token_id": 2,
+  "output_attentions": false,
+  "output_hidden_states": false,
+  "pad_token_id": 0,
+  "transformers_version": "5.12.1",
+  "use_cache": false
+}

model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2f9e0f3883b4b5c76643f2a3b67a86d049bf27c98e80b73a2bb9557a4e598217
+size 206783968

tinybrain_sft_meta.json

@@ -0,0 +1,10 @@
+{
+  "step": 522,
+  "val_loss": 2.2578155994415283,
+  "base_model": "exnivo/tinybrain-100m-base",
+  "dataset": "exnivo/tinybrain-instruct",
+  "batch_size": 100,
+  "block_size": 1024,
+  "epochs": 3,
+  "lr": 7e-05
+}

tokenizer_config.json

@@ -0,0 +1,22 @@
+{
+  "backend": "tokenizers",
+  "bos_token": "<|bos|>",
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<|eos|>",
+  "extra_special_tokens": [
+    "<|user|>",
+    "<|assistant|>",
+    "<|system|>",
+    "<|end|>",
+    "<|tool|>",
+    "<|tool_call|>",
+    "<|tool_result|>"
+  ],
+  "is_local": false,
+  "local_files_only": false,
+  "model_max_length": 2048,
+  "pad_token": "<|pad|>",
+  "padding_side": "right",
+  "tokenizer_class": "TokenizersBackend",
+  "truncation_side": "right"
+}