parameters guide
samplers guide
model generation
role play settings
quant selection
arm quants
iq quants vs q quants
optimal model setting
gibberish fixes
coherence
instructing following
quality generation
chat settings
quality settings
llamacpp server
llamacpp
lmstudio
sillytavern
koboldcpp
backyard
ollama
model generation steering
steering
model generation fixes
text generation webui
ggufs
exl2
full precision
quants
imatrix
neo imatrix
llama
llama-3
gemma
gemma2
gemma3
llama-2
llama-3.1
llama-3.2
mistral
Mixture of Experts
mixture of experts
mixtral
Update README.md
Browse files
README.md
CHANGED
|
@@ -274,8 +274,11 @@ Please see sections below this for advanced usage, more details, settings notes
|
|
| 274 |
| **Penalty Samplers** |
|
| 275 |
|
| 276 |
| repeat-last-n | Number of tokens to consider for penalties. Critical for preventing repetition. Default: 64 |
|
|
|
|
| 277 |
| repeat-penalty | Penalizes repeated token sequences. Range: 1.0-1.15. Default: 1.0 |
|
|
|
|
| 278 |
| presence-penalty | Penalizes token presence in previous text. Range: 0-0.2 for Class 3, 0.1-0.35 for Class 4 |
|
|
|
|
| 279 |
| frequency-penalty | Penalizes token frequency in previous text. Range: 0-0.25 for Class 3, 0.4-0.8 for Class 4 |
|
| 280 |
|
| 281 |
| penalize-nl | Penalizes newline tokens. Generally unused. Default: false |
|
|
@@ -284,10 +287,14 @@ Please see sections below this for advanced usage, more details, settings notes
|
|
| 284 |
| **Secondary Samplers** |
|
| 285 |
|
| 286 |
| mirostat | Controls perplexity during sampling. Modes: 0 (off), 1, or 2 |
|
|
|
|
| 287 |
| mirostat-lr | Mirostat learning rate. Default: 0.1 |
|
|
|
|
| 288 |
| mirostat-ent | Mirostat target entropy. Default: 5.0 |
|
| 289 |
|
|
|
|
| 290 |
| dynatemp-range | Range for dynamic temperature adjustment. Default: 0.0 |
|
|
|
|
| 291 |
| dynatemp-exp | Exponent for dynamic temperature scaling. Default: 1.0 |
|
| 292 |
|
| 293 |
| tfs | Tail free sampling - removes low-probability tokens. Default: 1.0 |
|
|
@@ -295,16 +302,20 @@ Please see sections below this for advanced usage, more details, settings notes
|
|
| 295 |
| typical | Selects tokens more likely than random given prior text. Default: 1.0 |
|
| 296 |
|
| 297 |
| xtc-probability | Probability of token removal. Range: 0-1 |
|
|
|
|
| 298 |
| xtc-threshold | Threshold for considering token removal. Default: 0.1 |
|
| 299 |
|
| 300 |
|
| 301 |
| **Advanced Samplers** |
|
| 302 |
|
| 303 |
| dry_multiplier | Controls DRY (Don't Repeat Yourself) intensity. Range: 0.8-1.12+ |
|
|
|
|
| 304 |
| dry_allowed_length | Allowed length for repeated sequences in DRY. Default: 2 |
|
|
|
|
| 305 |
| dry_base | Base value for DRY calculations. Range: 1.15-1.75+ for Class 4 |
|
| 306 |
|
| 307 |
| smoothing_factor | Quadratic sampling intensity. Range: 1-3 for Class 3, 3-5+ for Class 4 |
|
|
|
|
| 308 |
| smoothing_curve | Quadratic sampling curve. Range: 1 for Class 3, 1.5-2 for Class 4 |
|
| 309 |
|
| 310 |
|
|
|
|
| 274 |
| **Penalty Samplers** |
|
| 275 |
|
| 276 |
| repeat-last-n | Number of tokens to consider for penalties. Critical for preventing repetition. Default: 64 |
|
| 277 |
+
|
| 278 |
| repeat-penalty | Penalizes repeated token sequences. Range: 1.0-1.15. Default: 1.0 |
|
| 279 |
+
|
| 280 |
| presence-penalty | Penalizes token presence in previous text. Range: 0-0.2 for Class 3, 0.1-0.35 for Class 4 |
|
| 281 |
+
|
| 282 |
| frequency-penalty | Penalizes token frequency in previous text. Range: 0-0.25 for Class 3, 0.4-0.8 for Class 4 |
|
| 283 |
|
| 284 |
| penalize-nl | Penalizes newline tokens. Generally unused. Default: false |
|
|
|
|
| 287 |
| **Secondary Samplers** |
|
| 288 |
|
| 289 |
| mirostat | Controls perplexity during sampling. Modes: 0 (off), 1, or 2 |
|
| 290 |
+
|
| 291 |
| mirostat-lr | Mirostat learning rate. Default: 0.1 |
|
| 292 |
+
|
| 293 |
| mirostat-ent | Mirostat target entropy. Default: 5.0 |
|
| 294 |
|
| 295 |
+
|
| 296 |
| dynatemp-range | Range for dynamic temperature adjustment. Default: 0.0 |
|
| 297 |
+
|
| 298 |
| dynatemp-exp | Exponent for dynamic temperature scaling. Default: 1.0 |
|
| 299 |
|
| 300 |
| tfs | Tail free sampling - removes low-probability tokens. Default: 1.0 |
|
|
|
|
| 302 |
| typical | Selects tokens more likely than random given prior text. Default: 1.0 |
|
| 303 |
|
| 304 |
| xtc-probability | Probability of token removal. Range: 0-1 |
|
| 305 |
+
|
| 306 |
| xtc-threshold | Threshold for considering token removal. Default: 0.1 |
|
| 307 |
|
| 308 |
|
| 309 |
| **Advanced Samplers** |
|
| 310 |
|
| 311 |
| dry_multiplier | Controls DRY (Don't Repeat Yourself) intensity. Range: 0.8-1.12+ |
|
| 312 |
+
|
| 313 |
| dry_allowed_length | Allowed length for repeated sequences in DRY. Default: 2 |
|
| 314 |
+
|
| 315 |
| dry_base | Base value for DRY calculations. Range: 1.15-1.75+ for Class 4 |
|
| 316 |
|
| 317 |
| smoothing_factor | Quadratic sampling intensity. Range: 1-3 for Class 3, 3-5+ for Class 4 |
|
| 318 |
+
|
| 319 |
| smoothing_curve | Quadratic sampling curve. Range: 1 for Class 3, 1.5-2 for Class 4 |
|
| 320 |
|
| 321 |
|