upadted README and config

README.md

@@ -1,3 +1,104 @@
 ---
 license: apache-2.0
+pipeline_tag: text-generation
+library_name: transformers
+language: [
+    'am', 'ar', 'as', 'az', 'be', 'bg', 'bn', 'bn_rom', 'br',
+    'bs', 'ca', 'cs', 'cy', 'da', 'de', 'el', 'en', 'eo', 'es',
+    'et', 'eu', 'fa', 'ff', 'fi', 'fr', 'fy', 'ga', 'gd', 'gl',
+    'gn', 'gu', 'ha', 'he', 'hi', 'hi_rom', 'hr', 'ht', 'hu',
+    'hy', 'id', 'ig', 'is', 'it', 'ja', 'jv', 'ka', 'kk', 'km',
+    'kn', 'ko', 'ku', 'ky', 'la', 'lg', 'li', 'ln', 'lo', 'lt',
+    'lv', 'mg', 'mk', 'ml', 'mn', 'mr', 'ms', 'my', 'my_zaw',
+    'ne', 'nl', 'no', 'ns', 'om', 'or', 'pa', 'pl', 'ps', 'pt',
+    'qu', 'rm', 'ro', 'ru', 'sa', 'si', 'sc', 'sd', 'sk', 'sl',
+    'so', 'sq', 'sr', 'ss', 'su', 'sv', 'sw', 'ta', 'ta_rom',
+    'te', 'te_rom', 'th', 'tl', 'tn', 'tr', 'ug', 'uk', 'ur',
+    'ur_rom', 'uz', 'vi', 'wo', 'xh', 'yi', 'yo',
+    'zh-Hans', 'zh-Hant', 'zu',
+]
+datasets: [
+    'yahma/alpaca-cleaned',
+    'gbharti/wealth-alpaca_lora',
+    'databricks/databricks-dolly-15k',
+    'VMware/open-instruct',
+    'saillab/taco-datasets',
+    'xu-song/cc100-samples',
+    'jordiclive/wikipedia-summary-dataset',
+    'bigcode/the-stack-smol-xs',
+    'm-a-p/CodeFeedback-Filtered-Instruction',
+    'jtatman/python-code-dataset-500k',
+    'iamtarun/python_code_instructions_18k_alpaca',
+    'HuggingFaceH4/CodeAlpaca_20K',
+    'cognitivecomputations/dolphin-coder',
+    'fblgit/simple-math',
+    'gair-prox/open-web-math-pro',
+    'rvv-karma/Math-QA',
+    'ajibawa-2023/Maths-College',
+    'microsoft/orca-math-word-problems-200k',
+    'meta-math/MetaMathQA',
+    'TIGER-Lab/MathInstruct',
+    'TIGER-Lab/WebInstructSub',
+    'SkunkworksAI/reasoning-0.01',
+    'KingNish/reasoning-base-20k',
+    'Magpie-Align/Magpie-Reasoning-150K',
+    'thesven/gsm8k-reasoning',
+    'AlgorithmicResearchGroup/math_reasoning_autoformalization_track',
+    'badrex/llm-emoji-dataset',
+]
+tags:
+- litgpt
+- litdata
 ---
+
+# tangled-llama-t-32k-base-v0.1
+
+![logo](./misc/logo.png)
+
+A pretrained language model based on the Llama model with about **25M** parameters. This model has been trained on **22.1B** (`22,111,299,936`) tokens from more than **3.6M** (`3,597,088`) dataset rows.
+
+This model **isn't** designed for immediate use but rather for Continued Pretraining and Finetuning on a downstream task. While it can handle a context length of up to **128K** (`131,072`) tokens, it was pretrained with sequences of **2K** (`2048`) tokens.
+
+The objective is to streamline the cognitive or reasoning core, eliminating any redundant knowledge from the model.
+
+[loss, val_loss](https://api.wandb.ai/links/mtasic85/t66yvgeh)
+
+[val_ppl](https://api.wandb.ai/links/mtasic85/osr62qqd)
+
+[epoch](https://api.wandb.ai/links/mtasic85/pw0ilz5s)
+
+[learning_rate](https://api.wandb.ai/links/mtasic85/867ueoyx)
+
+## lm-evaluation-harness
+
+```bash
+litgpt evaluate --tasks 'hellaswag,gsm8k,truthfulqa_mc2,mmlu,winogrande,arc_challenge' --out_dir 'evaluate-quick/' --batch_size 4 --dtype 'bfloat16' out/pretrain/final/
+```
+
+```bash
+litgpt evaluate --tasks 'leaderboard' --out_dir 'evaluate-leaderboard/' --batch_size 4 --dtype 'bfloat16' out/pretrain/final/
+```
+
+```bash
+litgpt evaluate --tasks 'bbh_zeroshot,bbh_fewshot,bbh_cot_fewshot,bbh_cot_zeroshot' --out_dir 'evaluate-bigbenchhard/' --batch_size 4 --dtype 'bfloat16' out/pretrain/final/
+```
+
+```bash
+litgpt evaluate --tasks 'mmlu,mmlu_pro' --out_dir 'evaluate-mmlu/' --batch_size 4 --dtype 'bfloat16' out/pretrain/final/
+```
+
+```bash
+litgpt evaluate --tasks 'arc_challenge,boolq,gpqa,hellaswag,openbookqa,piqa,truthfulqa_mc2,winogrande' --out_dir 'evaluate-reasoning/' --batch_size 4 --dtype 'bfloat16' out/pretrain/final/
+```
+
+```bash
+litgpt evaluate --tasks 'mmlu_multilingual,mgsm' --out_dir 'evaluate-multilinguals/' --batch_size 4 --dtype 'bfloat16' out/pretrain/final/
+```
+
+```bash
+litgpt evaluate --tasks 'gsm8k,mathqa' --out_dir 'evaluate-math/' --batch_size 4 --dtype 'bfloat16' out/pretrain/final/
+```
+
+```bash
+litgpt evaluate --tasks 'wikitext,qasper' --out_dir 'evaluate-long/' --batch_size 4 --dtype 'bfloat16' out/pretrain/final/
+```

config.json

@@ -1,16 +1,16 @@
 {
-  "_name_or_path": "tangledgroup/tangled-llama-33m-32k-base-v0.1",
+  "_name_or_path": "tangledgroup/tangled-llama-t-128k-base-v0.1",
   "architectures": [
     "LlamaForCausalLM"
   ],
   "bos_token_id": 1,
   "eos_token_id": 2,
-  "hidden_size": 1024,
-  "intermediate_size": 4096,
+  "hidden_size": 256,
+  "intermediate_size": 1024,
   "max_position_embeddings": 38400,
   "model_type": "llama",
   "num_attention_heads": 32,
-  "num_hidden_layers": 5,
+  "num_hidden_layers": 16,
   "num_key_value_heads": 8,
   "rms_norm_eps": 1e-05,
   "rope_scaling": null,