Archaeo-32B-KTO / README.md
Delta-Vector's picture
Update README.md
ececa61 verified
---
tags:
- roleplay
- creative-writing
- merge
- mergekit
- kto
- rl
- finetune
datasets:
- Delta-Vector/Tauri-IFeval-Dans-Tulu-KTO
- Delta-Vector/Tauri-Opus-Accepted-GPT-Rejected-Opus-Writing-Prompts
- Delta-Vector/Tauri-KTO-Instruct-Mix
- Delta-Vector/Tauri-Purpura-Arkhaios-CC-KTO
base_model:
- Delta-Vector/Archaeo-32B
pipeline_tag: text-generation
library_name: transformers
---
```
__~a~_
~~; ~_
_ ~ ~_ _
'_\;__._._._._._._] ~_._._._._._.__;/_`
'(/'/'/'/'|'|'|'| ( )|'|'|'|'\'\'\'\)'
(/ / / /, | | | |(/ \) | | | ,\ \ \ \)
(/ / / / / | | | ~(/ \) ~ | | \ \ \ \ \)
(/ / / / / ~ ~ ~ (/ \) ~ ~ \ \ \ \ \)
(/ / / / ~ / (||)| ~ \ \ \ \)
~ / / ~ M /||\M ~ \ \ ~
~ ~ /||\ ~ ~
//||\\
//||\\
//||\\
'/||\' "Archaeopteryx"
```
<html lang="en">
<head>
<meta charset="UTF-8">
<meta name="viewport" content="width=device-width, initial-scale=1.0">
<style>
@import url('https://fonts.googleapis.com/css2?family=VT323&display=swap');
body {
background: #0a0017;
margin: 0;
padding: 20px;
font-family: 'VT323', monospace;
color: #ff00aa;
text-shadow: 0 0 8px #ff00aa;
animation: glitch-flicker 0.2s infinite alternate;
}
@keyframes glitch-flicker {
0% { text-shadow: 0 0 5px #ff00aa, 0 0 15px #ff00aa; }
100% { text-shadow: 0 0 8px #ff0066, 0 0 18px #ff0066; }
}
.crt-container {
padding: 10px;
max-width: 900px;
margin: auto;
}
.crt-case {
background: linear-gradient(135deg, #130021, #20002c);
border-radius: 10px;
padding: 15px;
box-shadow:
inset 2px 2px 10px rgba(255,0,170,0.5),
2px 2px 5px rgba(255,0,170,0.3),
0 0 25px rgba(255,0,170,0.2);
}
.crt-screen {
background: #0c011a;
padding: 20px;
border-radius: 10px;
box-shadow:
inset 0 0 25px rgba(255,0,170,0.3),
0 0 15px rgba(255,0,170,0.7);
filter: contrast(1.2) brightness(1.2);
text-shadow: 0px 0px 5px #ff00aa;
animation: glow-pulse 3s infinite alternate;
}
@keyframes glow-pulse {
0% { box-shadow: inset 0 0 20px rgba(255,0,170,0.3), 0 0 15px rgba(255,0,170,0.3); }
100% { box-shadow: inset 0 0 30px rgba(255,0,170,0.5), 0 0 25px rgba(255,0,170,0.5); }
}
h2 {
color: #ff33cc;
text-align: center;
font-size: 28px;
text-shadow:
0 0 8px #ff33cc,
0 0 18px #ff0044;
}
pre {
background: rgba(255,0,170,0.1);
padding: 10px;
border-radius: 10px;
color: #ff66cc;
font-size: 14px;
box-shadow: inset 0 0 10px rgba(255,0,170,0.5);
}
.glitch {
animation: text-glitch 0.5s infinite alternate;
}
@keyframes text-glitch {
0% { transform: translateX(-2px); text-shadow: 0 0 5px #ff0066, 0 0 10px #ff33cc; }
100% { transform: translateX(2px); text-shadow: 0 0 8px #ff00aa, 0 0 20px #ff0099; }
}
.neon-link {
color: #ff66cc;
text-decoration: none;
transition: text-shadow 0.3s ease;
}
.neon-link:hover {
text-shadow: 0px 0px 15px #ff66cc, 0 0 25px rgba(255,0,170,0.5);
}
.ascii-art {
text-align: center;
font-size: 12px;
color: #ff33cc;
text-shadow: 0px 0px 5px #ff00ff;
margin-bottom: 20px;
}
.quantso-container {
display: flex;
justify-content: center;
gap: 20px;
margin-top: 20px;
}
.quantso-box {
background: rgba(255,0,170,0.1);
padding: 15px;
border-radius: 10px;
text-align: center;
box-shadow: inset 0 0 10px rgba(255,0,170,0.5);
flex: 1;
max-width: 150px;
}
</style>
</head>
<body>
<div class="crt-container">
<div class="crt-case">
<div class="crt-screen">
<p>Support me on Ko-Fi: https://ko-fi.com/deltavector</p>
<p>A series of Merges made for Roleplaying & Creative Writing, This model is a RL train ontop of Archaeo. A merge using Hamanasu-Magnum & Kunou, Trained with Axolotl on 8xH200s.</p>
<h3>ChatML formatting</h3>
<pre>
"""<|im_start|>system
system prompt<|im_end|>
<|im_start|>user
Hi there!<|im_end|>
<|im_start|>assistant
Nice to meet you!<|im_end|>
<|im_start|>user
Can I ask a question?<|im_end|>
<|im_start|>assistant
"""
</pre>
<h3>Axolotl Configuration</h3>
<pre>
base_model: ./model
plugins:
- axolotl.integrations.liger.LigerPlugin
- axolotl.integrations.cut_cross_entropy.CutCrossEntropyPlugin
liger_rope: true
liger_rms_norm: true
liger_layer_norm: true
liger_glu_activation: true
liger_fused_linear_cross_entropy: true
cut_cross_entropy: false
load_in_8bit: false
load_in_4bit: false
strict: false
rl: kto
kto_undesirable_weight: 1.0
datasets:
- path: Delta-Vector/Tauri-Opus-Accepted-GPT-Rejected-Opus-Writing-Prompts
split: train
type: chatml.argilla
- path: Delta-Vector/Tauri-IFeval-Dans-Tulu-KTO
split: train
type: chatml.argilla
- path: Delta-Vector/Tauri-KTO-Instruct-Mix
split: train
type: chatml.argilla
- path: Delta-Vector/Tauri-Purpura-Arkhaios-CC-KTO
split: train
type: chatml.argilla
dataset_prepared_path: last_run_prepared
val_set_size: 0.0
output_dir: ./archaeo-kto-v2
remove_unused_columns: false
#@lora_mlp_kernel: true
#lora_qkv_kernel: true
#lora_o_kernel: true
adapter: lora
lora_model_dir:
sequence_len: 8192
pad_to_sequence_len: false
lora_r: 64
lora_alpha: 32
lora_dropout: 0.0
lora_target_linear: true
lora_fan_in_fan_out:
lora_target_modules:
- gate_proj
- down_proj
- up_proj
- q_proj
- v_proj
- k_proj
- o_proj
wandb_project: Francois-V2
wandb_entity:
wandb_watch:
wandb_name: Archaeo-32b-KTO
wandb_log_model:
gradient_accumulation_steps: 4
micro_batch_size: 4
num_epochs: 1
optimizer: paged_ademamix_8bit
lr_scheduler: constant_with_warmup
learning_rate: 5e-6
max_grad_norm: 0.001
train_on_inputs: false
group_by_length: false
bf16: auto
fp16:
tf32: true
gradient_checkpointing: true
gradient_checkpointing_kwargs:
use_reentrant: true
early_stopping_patience:
resume_from_checkpoint:
local_rank:
logging_steps: 1
xformers_attention:
flash_attention: true
warmup_steps: 100
evals_per_epoch: 4
eval_table_size:
eval_max_new_tokens: 128
saves_per_epoch: 1
debug:
deepspeed: ./deepspeed_configs/zero3_bf16.json
weight_decay: 0.0025
fsdp:
fsdp_config:
</pre>
<h3>Quants:</h3>
<div class="quantso-container">
<div class="quantso-box">
<strong>GGUF</strong><br>
<a class="neon-link" href="#">https://huggingface.co/mradermacher/Axo-Merge-Archaeo-V2-Lora-GGUF</a>
</div>
<div class="quantso-box">
<strong>EXL2</strong><br>
<a class="neon-link" href="#">Delta-Vector/Archaeo-32B-KTO-exl2</a>
</div>
</div>
<h3>Credits</h3>
<p>Thank you to: Kubernetes-bad, LucyKnada, Kalomaze, Alicat, Intervitens, Samantha Twinkman, Tav, Trappu & The rest of Anthracite</p>
</div>
</div>
</div>
</body>
</html>