Kyllene-34B-v1.1-5.0bpw-h6-exl2 / mergemonster_kyllene_v11.txt
LoneStriker's picture
Upload folder using huggingface_hub
22f70c4 verified
⠀⠀⠀⠀⠀⠀⣀⡀⠀⠀⣀⣤⣶⣾⣿⣿⣷⣶⣤⣀⠀⠀⣀⣀⠀⠀⠀⠀⠀⠀
⠀⠀⠀⠀⠀⠜⠉⣿⡆⣼⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣧⢰⣿⠉⠃⠀⠀⠀⠀⠀
⠀⢀⣤⣴⣦⣄⣴⠟⣸⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⡎⢻⣦⣠⣴⣦⣄⠀⠀
⠀⡞⠁⣠⣾⢿⣧⠀⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⠀⣽⡿⣷⣄⠈⢷⠀
⠀⣠⣾⠟⠁⢸⣿⠀⠘⢿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⡿⠁⠀⣿⡇⠈⠻⣷⣄⠀
⣰⡿⠁⠀⢀⣾⣏⣾⣄⣰⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣇⣰⣷⣹⣷⠀⠀⠈⢿⣆
⣿⡇⠀⢠⣾⠏⢸⣿⣿⣿⣿⠋⢻⣿⣿⣿⣿⡟⠙⣿⣿⣿⣿⡇⠹⣷⡀⠀⢸⣿
⠹⣿⣴⡿⠋⠀⠈⠛⠉⣹⣿⣦⣄⡹⣿⣿⣋⣠⣶⣿⣏⠉⠛⠁⠀⠙⢿⣦⣿⠏
⠀⣸⣿⠿⠿⣿⣾⣿⡿⠿⣿⣿⣿⣿⡆⢰⣿⣿⣿⣿⠿⢿⣿⣶⣿⠿⠿⣻⣇⠀
⠀⣿⡇⢀⣴⣶⣤⣀⣴⣿⠿⣻⡿⣿⣧⣾⣿⢿⣟⠿⣿⣦⣀⣤⣶⣦⠀⢸⣿⠀
⠀⢿⣧⠈⠃⢀⣵⣿⡋⠁⢀⣿⡷⣿⡇⢻⣿⣿⣿⡀⠈⢛⣿⣮⡀⠘⠀⣼⡟⠀
⠀⠈⠻⣷⣤⣟⣋⣿⣧⣴⡿⠋⠀⣿⡇⢸⣿⠀⠙⢿⣦⣼⣿⣙⣻⣤⣾⠟⠁⠀
⠀⠀⠀⠈⢽⣿⠛⢻⣏⢉⣤⣶⣶⣿⠁⠈⣿⣶⣶⣤⡉⣽⡟⠛⣿⡏⠁⠀⠀⠀
⠀⠀⠀⠀⠈⠿⣷⣾⣾⣟⣉⣠⣿⢿⡇⢸⠿⣿⣄⣙⣻⣷⣷⣾⠿⠁⠀⠀⠀⠀
⠀⠀⠀⠀⠀⠀⠀⠀⠙⠻⠿⠛⢁⡼⠃⠘⢦⡈⠛⠿⠟⠃⠀⠀⠀⠀⠀⠀⠀⠀
01:05:33 - THE MERGE MONSTER HUNGERS
------------------------------------
Device : cpu
Random seed : 42
Starting model : ../jondurbin_bagel-dpo-34b-v0.2
Models to merge : ['../NousResearch_Nous-Capybara-34B', '../NousResearch_Nous-Hermes-2-Yi-34B', '../SUSTech_SUS-Chat-34B']
Output directory : ./mm-output
Phrases loaded : 31
Auto weights : False
Merge ratios : [0.2, 0.4, 0.6, 0.8]
Merge method(s) : ['slerp']
Merge headers : True
Strategy used : cumulative
------------------------------------
01:05:34 - Loading model (../jondurbin_bagel-dpo-34b-v0.2)...
Loading checkpoint shards: 100%|████████████████| 15/15 [00:32<00:00, 2.18s/it]
01:06:59 - Model loaded. Dtype: torch.float16
------------------------------------
-----------------------------------------------------------------------------------------------------
| Type | Phrase | Context | Raw Prob* | Used Prob** | Change |
-----------------------------------------------------------------------------------------------------
| BAD | anticipation | Her body quivers with | 0.00000% | 0.00% | N/A |
| BAD | anticipation | The atmosphere is thic.. | 0.00000% | 0.00% | N/A |
| BAD | unwavering | Filled with an | 0.00000% | 0.00% | N/A |
| BAD | determination | Her eyes were filled w.. | 0.00000% | 0.00% | N/A |
| BAD | determination | Her stubbornness only .. | 0.00000% | 0.00% | N/A |
| BAD | whisper | Her voice barely above.. | 0.00000% | 0.00% | N/A |
| BAD | spine | shivers down her | 0.00000% | 0.00% | N/A |
| BAD | sends shivers | The thrill of the act | 0.00000% | 0.00% | N/A |
| BAD | ministrations | She moans and twitches.. | 0.00006% | 0.00% | N/A |
| BAD | legs | wraps her | 0.00000% | 0.00% | N/A |
| BAD | imposing figure | He had an | 0.00000% | 0.00% | N/A |
| BAD | shared challenges | Their bond strengthene.. | 0.00001% | 0.00% | N/A |
| BAD | bond | forged a | 0.00008% | 0.00% | N/A |
| BAD | bond | an unspoken | 0.00009% | 0.00% | N/A |
| BAD | enhance our expe.. | I'm excited to see how | 0.00000% | 0.00% | N/A |
| BAD | sense of vulnera.. | create a | 0.00000% | 0.00% | N/A |
| BAD | dimensions of in.. | explore new | 0.00000% | 0.00% | N/A |
| BAD | deepening our co.. | while | 0.00000% | 0.00% | N/A |
| BAD | shared experiences | through | 0.00000% | 0.00% | N/A |
| BAD | societal expecta.. | that transcend | 0.00000% | 0.00% | N/A |
| BAD | conventional bou.. | that defy | 0.00000% | 0.00% | N/A |
| BAD | conventional bou.. | and defy | 0.00000% | 0.00% | N/A |
| BAD | open communication | an environment | 0.00000% | 0.00% | N/A |
| BAD | emotional vulner.. | an environment | 0.00000% | 0.00% | N/A |
| BAD | heightens our co.. | touch and the anticipa.. | 0.00000% | 0.00% | N/A |
| BAD | sensations you'r.. | I'm enjoying | 0.00000% | 0.00% | N/A |
| BAD | is truly arousing | attention to detail | 0.00000% | 0.00% | N/A |
| BAD | is truly arousing | way you explore my body | 0.00000% | 0.00% | N/A |
| BAD | challenge presen.. | my resolve unwavering .. | 0.00000% | 0.00% | N/A |
| BAD | humble vessel | surrendering to the ex.. | 0.00000% | 0.00% | N/A |
| BAD | bond | cherishing the unique | 0.00013% | 0.00% | N/A |
| BAD | bond | special | 0.00030% | 0.00% | N/A |
| BAD | grows stronger w.. | bond | 0.00000% | 0.00% | N/A |
| BAD | that cannot be b.. | bond | 0.00000% | 0.00% | N/A |
| BAD | becomes unbreaka.. | bond | 0.00000% | 0.00% | N/A |
| BAD | grew stronger wi.. | bond | 0.00000% | 0.00% | N/A |
| GOOD | The apple is in .. | Question: If I'm in th.. | 0.00139% | 0.00% | N/A |
------------------------------------------------------------------------------------------------------
| Totals | 0.00% | 0.01% | 0.00% |
------------------------------------------------------------------------------------------------------
* = Unweighted, raw probability - ** = Probability after weight adjustments
------------------------------------
01:07:39 - Loading model (../NousResearch_Nous-Capybara-34B)...
Loading checkpoint shards: 100%|██████████████████| 7/7 [01:04<00:00, 9.19s/it]
01:09:33 - Model loaded. Dtype: torch.float16
------------------------------------
Optimizing Layer 1/60 (slerp): 100%|██████████████| 4/4 [04:01<00:00, 60.38s/it]
[[1.0, 'jondurbin_bagel-dpo-34b-v0.2'], [0.2, 'NousResearch_Nous-Capybara-34B']]
01:15:02 - Layer 1/60 - CHANGED - 0.00007 > 0.00006 - 2.5%
----
Optimizing Layer 2/60 (slerp): 100%|██████████████| 4/4 [03:52<00:00, 58.04s/it]
[[1.0, 'jondurbin_bagel-dpo-34b-v0.2'], [0.4, 'NousResearch_Nous-Capybara-34B']]
01:20:22 - Layer 2/60 - CHANGED - 0.00006 > 0.00006 - 1.6%
----
Optimizing Layer 3/60 (slerp): 100%|██████████████| 4/4 [04:03<00:00, 60.90s/it]
[[1.0, 'jondurbin_bagel-dpo-34b-v0.2']]
01:25:50 - Layer 3/60 - RETAINED - 0.00006
----
Optimizing Layer 4/60 (slerp): 100%|██████████████| 4/4 [05:28<00:00, 82.25s/it]
[[1.0, 'jondurbin_bagel-dpo-34b-v0.2']]
01:32:54 - Layer 4/60 - RETAINED - 0.00006
----
Optimizing Layer 5/60 (slerp): 100%|██████████████| 4/4 [04:15<00:00, 63.94s/it]
[[1.0, 'jondurbin_bagel-dpo-34b-v0.2']]
01:38:53 - Layer 5/60 - RETAINED - 0.00006
----
Optimizing Layer 6/60 (slerp): 100%|██████████████| 4/4 [04:16<00:00, 64.24s/it]
[[1.0, 'jondurbin_bagel-dpo-34b-v0.2']]
01:44:47 - Layer 6/60 - RETAINED - 0.00006
----
Optimizing Layer 7/60 (slerp): 100%|██████████████| 4/4 [04:04<00:00, 61.02s/it]
[[1.0, 'jondurbin_bagel-dpo-34b-v0.2']]
01:50:20 - Layer 7/60 - RETAINED - 0.00006
----
Optimizing Layer 8/60 (slerp): 100%|██████████████| 4/4 [04:07<00:00, 61.95s/it]
[[1.0, 'jondurbin_bagel-dpo-34b-v0.2']]
01:55:59 - Layer 8/60 - RETAINED - 0.00006
----
Optimizing Layer 9/60 (slerp): 100%|██████████████| 4/4 [04:04<00:00, 61.17s/it]
[[1.0, 'jondurbin_bagel-dpo-34b-v0.2'], [0.4, 'NousResearch_Nous-Capybara-34B']]
02:01:26 - Layer 9/60 - CHANGED - 0.00006 > 0.00006 - 1.3%
----
Optimizing Layer 10/60 (slerp): 100%|█████████████| 4/4 [03:56<00:00, 59.05s/it]
[[1.0, 'jondurbin_bagel-dpo-34b-v0.2']]
02:06:41 - Layer 10/60 - RETAINED - 0.00006
----
Optimizing Layer 11/60 (slerp): 100%|█████████████| 4/4 [03:43<00:00, 55.90s/it]
[[1.0, 'jondurbin_bagel-dpo-34b-v0.2'], [0.8, 'NousResearch_Nous-Capybara-34B']]
02:11:45 - Layer 11/60 - CHANGED - 0.00006 > 0.00006 - 4.8%
----
Optimizing Layer 12/60 (slerp): 100%|█████████████| 4/4 [03:53<00:00, 58.32s/it]
[[1.0, 'jondurbin_bagel-dpo-34b-v0.2'], [0.8, 'NousResearch_Nous-Capybara-34B']]
02:16:54 - Layer 12/60 - CHANGED - 0.00006 > 0.00005 - 12.2%
----
Optimizing Layer 13/60 (slerp): 100%|█████████████| 4/4 [04:09<00:00, 62.31s/it]
[[1.0, 'jondurbin_bagel-dpo-34b-v0.2'], [0.4, 'NousResearch_Nous-Capybara-34B']]
02:22:31 - Layer 13/60 - CHANGED - 0.00005 > 0.00005 - 3.6%
----
Optimizing Layer 14/60 (slerp): 100%|█████████████| 4/4 [03:31<00:00, 52.84s/it]
[[1.0, 'jondurbin_bagel-dpo-34b-v0.2'], [0.4, 'NousResearch_Nous-Capybara-34B']]
02:27:20 - Layer 14/60 - CHANGED - 0.00005 > 0.00005 - 1.5%
----
Optimizing Layer 15/60 (slerp): 100%|█████████████| 4/4 [04:26<00:00, 66.67s/it]
[[1.0, 'jondurbin_bagel-dpo-34b-v0.2']]
02:33:32 - Layer 15/60 - RETAINED - 0.00005
----
Optimizing Layer 16/60 (slerp): 100%|█████████████| 4/4 [04:36<00:00, 69.09s/it]
[[1.0, 'jondurbin_bagel-dpo-34b-v0.2']]
02:39:38 - Layer 16/60 - RETAINED - 0.00005
----
Optimizing Layer 17/60 (slerp): 100%|█████████████| 4/4 [04:22<00:00, 65.64s/it]
[[1.0, 'jondurbin_bagel-dpo-34b-v0.2']]
02:45:41 - Layer 17/60 - RETAINED - 0.00005
----
Optimizing Layer 18/60 (slerp): 100%|█████████████| 4/4 [04:39<00:00, 69.87s/it]
[[1.0, 'jondurbin_bagel-dpo-34b-v0.2']]
02:51:51 - Layer 18/60 - RETAINED - 0.00005
----
Optimizing Layer 19/60 (slerp): 100%|█████████████| 4/4 [05:10<00:00, 77.56s/it]
[[1.0, 'jondurbin_bagel-dpo-34b-v0.2']]
02:58:36 - Layer 19/60 - RETAINED - 0.00005
----
Optimizing Layer 20/60 (slerp): 100%|█████████████| 4/4 [05:03<00:00, 75.87s/it]
[[1.0, 'jondurbin_bagel-dpo-34b-v0.2'], [0.8, 'NousResearch_Nous-Capybara-34B']]
03:05:16 - Layer 20/60 - CHANGED - 0.00005 > 0.00005 - 0.2%
----
Optimizing Layer 21/60 (slerp): 100%|█████████████| 4/4 [05:42<00:00, 85.60s/it]
[[1.0, 'jondurbin_bagel-dpo-34b-v0.2'], [0.8, 'NousResearch_Nous-Capybara-34B']]
03:12:46 - Layer 21/60 - CHANGED - 0.00005 > 0.00001 - 77.3%
----
Optimizing Layer 22/60 (slerp): 100%|█████████████| 4/4 [05:48<00:00, 87.20s/it]
[[1.0, 'jondurbin_bagel-dpo-34b-v0.2'], [0.8, 'NousResearch_Nous-Capybara-34B']]
03:21:02 - Layer 22/60 - CHANGED - 0.00001 > -0.00000 - 126.4%
----
Optimizing Layer 23/60 (slerp): 100%|████████████| 4/4 [07:03<00:00, 105.79s/it]
[[1.0, 'jondurbin_bagel-dpo-34b-v0.2'], [0.8, 'NousResearch_Nous-Capybara-34B']]
03:30:53 - Layer 23/60 - CHANGED - -0.00000 > -0.00003 - 988.2%
----
Optimizing Layer 24/60 (slerp): 100%|█████████████| 4/4 [06:11<00:00, 92.99s/it]
[[1.0, 'jondurbin_bagel-dpo-34b-v0.2'], [0.8, 'NousResearch_Nous-Capybara-34B']]
03:39:09 - Layer 24/60 - CHANGED - -0.00003 > -0.00006 - 90.8%
----
Optimizing Layer 25/60 (slerp): 100%|█████████████| 4/4 [05:42<00:00, 85.51s/it]
[[1.0, 'jondurbin_bagel-dpo-34b-v0.2'], [0.8, 'NousResearch_Nous-Capybara-34B']]
03:46:40 - Layer 25/60 - CHANGED - -0.00006 > -0.00013 - 105.5%
----
Optimizing Layer 26/60 (slerp): 100%|█████████████| 4/4 [05:10<00:00, 77.58s/it]
[[1.0, 'jondurbin_bagel-dpo-34b-v0.2'], [0.4, 'NousResearch_Nous-Capybara-34B']]
03:53:21 - Layer 26/60 - CHANGED - -0.00013 > -0.00014 - 8.8%
----
Optimizing Layer 27/60 (slerp): 100%|█████████████| 4/4 [04:53<00:00, 73.48s/it]
[[1.0, 'jondurbin_bagel-dpo-34b-v0.2']]
03:59:41 - Layer 27/60 - RETAINED - -0.00014
----
Optimizing Layer 28/60 (slerp): 100%|█████████████| 4/4 [05:00<00:00, 75.07s/it]
[[1.0, 'jondurbin_bagel-dpo-34b-v0.2'], [0.4, 'NousResearch_Nous-Capybara-34B']]
04:06:11 - Layer 28/60 - CHANGED - -0.00014 > -0.00015 - 9.9%
----
Optimizing Layer 29/60 (slerp): 100%|█████████████| 4/4 [05:18<00:00, 79.66s/it]
[[1.0, 'jondurbin_bagel-dpo-34b-v0.2'], [0.8, 'NousResearch_Nous-Capybara-34B']]
04:13:06 - Layer 29/60 - CHANGED - -0.00015 > -0.00026 - 73.9%
----
Optimizing Layer 30/60 (slerp): 100%|█████████████| 4/4 [04:39<00:00, 69.97s/it]
[[1.0, 'jondurbin_bagel-dpo-34b-v0.2'], [0.2, 'NousResearch_Nous-Capybara-34B']]
04:19:19 - Layer 30/60 - CHANGED - -0.00026 > -0.00026 - 0.1%
----
Optimizing Layer 31/60 (slerp): 100%|█████████████| 4/4 [05:03<00:00, 75.98s/it]
[[1.0, 'jondurbin_bagel-dpo-34b-v0.2'], [0.8, 'NousResearch_Nous-Capybara-34B']]
04:26:15 - Layer 31/60 - CHANGED - -0.00026 > -0.00045 - 73.2%
----
Optimizing Layer 32/60 (slerp): 100%|█████████████| 4/4 [04:50<00:00, 72.61s/it]
[[1.0, 'jondurbin_bagel-dpo-34b-v0.2']]
04:32:41 - Layer 32/60 - RETAINED - -0.00045
----
Optimizing Layer 33/60 (slerp): 100%|█████████████| 4/4 [04:42<00:00, 70.72s/it]
[[1.0, 'jondurbin_bagel-dpo-34b-v0.2']]
04:38:55 - Layer 33/60 - RETAINED - -0.00045
----
Optimizing Layer 34/60 (slerp): 100%|█████████████| 4/4 [05:10<00:00, 77.62s/it]
[[1.0, 'jondurbin_bagel-dpo-34b-v0.2']]
04:45:43 - Layer 34/60 - RETAINED - -0.00045
----
Optimizing Layer 35/60 (slerp): 100%|█████████████| 4/4 [05:18<00:00, 79.62s/it]
[[1.0, 'jondurbin_bagel-dpo-34b-v0.2']]
04:52:33 - Layer 35/60 - RETAINED - -0.00045
----
Optimizing Layer 36/60 (slerp): 100%|█████████████| 4/4 [05:31<00:00, 82.80s/it]
[[1.0, 'jondurbin_bagel-dpo-34b-v0.2'], [0.8, 'NousResearch_Nous-Capybara-34B']]
04:59:39 - Layer 36/60 - CHANGED - -0.00045 > -0.00058 - 27.3%
----
Optimizing Layer 37/60 (slerp): 100%|█████████████| 4/4 [05:40<00:00, 85.08s/it]
[[1.0, 'jondurbin_bagel-dpo-34b-v0.2'], [0.8, 'NousResearch_Nous-Capybara-34B']]
05:07:00 - Layer 37/60 - CHANGED - -0.00058 > -0.00068 - 17.0%
----
Optimizing Layer 38/60 (slerp): 100%|█████████████| 4/4 [05:09<00:00, 77.43s/it]
[[1.0, 'jondurbin_bagel-dpo-34b-v0.2']]
05:13:50 - Layer 38/60 - RETAINED - -0.00068
----
Optimizing Layer 39/60 (slerp): 100%|█████████████| 4/4 [04:52<00:00, 73.15s/it]
[[1.0, 'jondurbin_bagel-dpo-34b-v0.2'], [0.8, 'NousResearch_Nous-Capybara-34B']]
05:20:23 - Layer 39/60 - CHANGED - -0.00068 > -0.00094 - 38.6%
----
Optimizing Layer 40/60 (slerp): 100%|█████████████| 4/4 [05:11<00:00, 77.87s/it]
[[1.0, 'jondurbin_bagel-dpo-34b-v0.2']]
05:27:10 - Layer 40/60 - RETAINED - -0.00094
----
Optimizing Layer 41/60 (slerp): 100%|█████████████| 4/4 [04:56<00:00, 74.02s/it]
[[1.0, 'jondurbin_bagel-dpo-34b-v0.2']]
05:33:43 - Layer 41/60 - RETAINED - -0.00094
----
Optimizing Layer 42/60 (slerp): 100%|█████████████| 4/4 [05:11<00:00, 77.90s/it]
[[1.0, 'jondurbin_bagel-dpo-34b-v0.2']]
05:40:32 - Layer 42/60 - RETAINED - -0.00094
----
Optimizing Layer 43/60 (slerp): 100%|█████████████| 4/4 [05:07<00:00, 76.91s/it]
[[1.0, 'jondurbin_bagel-dpo-34b-v0.2']]
05:47:21 - Layer 43/60 - RETAINED - -0.00094
----
Optimizing Layer 44/60 (slerp): 100%|█████████████| 4/4 [05:27<00:00, 81.99s/it]
[[1.0, 'jondurbin_bagel-dpo-34b-v0.2']]
05:54:34 - Layer 44/60 - RETAINED - -0.00094
----
Optimizing Layer 45/60 (slerp): 100%|█████████████| 4/4 [05:55<00:00, 88.94s/it]
[[1.0, 'jondurbin_bagel-dpo-34b-v0.2']]
06:02:20 - Layer 45/60 - RETAINED - -0.00094
----
Optimizing Layer 46/60 (slerp): 100%|█████████████| 4/4 [05:23<00:00, 80.84s/it]
[[1.0, 'jondurbin_bagel-dpo-34b-v0.2']]
06:09:36 - Layer 46/60 - RETAINED - -0.00094
----
Optimizing Layer 47/60 (slerp): 100%|█████████████| 4/4 [05:10<00:00, 77.74s/it]
[[1.0, 'jondurbin_bagel-dpo-34b-v0.2']]
06:16:33 - Layer 47/60 - RETAINED - -0.00094
----
Optimizing Layer 48/60 (slerp): 100%|█████████████| 4/4 [04:53<00:00, 73.39s/it]
[[1.0, 'jondurbin_bagel-dpo-34b-v0.2']]
06:23:12 - Layer 48/60 - RETAINED - -0.00094
----
Optimizing Layer 49/60 (slerp): 100%|█████████████| 4/4 [05:12<00:00, 78.19s/it]
[[1.0, 'jondurbin_bagel-dpo-34b-v0.2'], [0.4, 'NousResearch_Nous-Capybara-34B']]
06:30:19 - Layer 49/60 - CHANGED - -0.00094 > -0.00100 - 6.8%
----
Optimizing Layer 50/60 (slerp): 100%|█████████████| 4/4 [05:16<00:00, 79.20s/it]
[[1.0, 'jondurbin_bagel-dpo-34b-v0.2'], [0.6, 'NousResearch_Nous-Capybara-34B']]
06:37:19 - Layer 50/60 - CHANGED - -0.00100 > -0.00106 - 6.1%
----
Optimizing Layer 51/60 (slerp): 100%|█████████████| 4/4 [05:08<00:00, 77.05s/it]
[[1.0, 'jondurbin_bagel-dpo-34b-v0.2']]
06:44:03 - Layer 51/60 - RETAINED - -0.00106
----
Optimizing Layer 52/60 (slerp): 100%|█████████████| 4/4 [04:41<00:00, 70.42s/it]
[[1.0, 'jondurbin_bagel-dpo-34b-v0.2'], [0.8, 'NousResearch_Nous-Capybara-34B']]
06:50:20 - Layer 52/60 - CHANGED - -0.00106 > -0.00128 - 20.3%
----
Optimizing Layer 53/60 (slerp): 100%|█████████████| 4/4 [05:05<00:00, 76.48s/it]
[[1.0, 'jondurbin_bagel-dpo-34b-v0.2'], [0.2, 'NousResearch_Nous-Capybara-34B']]
06:57:10 - Layer 53/60 - CHANGED - -0.00128 > -0.00128 - 0.2%
----
Optimizing Layer 54/60 (slerp): 100%|█████████████| 4/4 [05:37<00:00, 84.37s/it]
[[1.0, 'jondurbin_bagel-dpo-34b-v0.2'], [0.4, 'NousResearch_Nous-Capybara-34B']]
07:04:24 - Layer 54/60 - CHANGED - -0.00128 > -0.00132 - 3.5%
----
Optimizing Layer 55/60 (slerp): 100%|█████████████| 4/4 [06:07<00:00, 91.86s/it]
[[1.0, 'jondurbin_bagel-dpo-34b-v0.2']]
07:12:17 - Layer 55/60 - RETAINED - -0.00132
----
Optimizing Layer 56/60 (slerp): 100%|█████████████| 4/4 [05:23<00:00, 80.92s/it]
[[1.0, 'jondurbin_bagel-dpo-34b-v0.2'], [0.8, 'NousResearch_Nous-Capybara-34B']]
07:19:47 - Layer 56/60 - CHANGED - -0.00132 > -0.00152 - 14.7%
----
Optimizing Layer 57/60 (slerp): 100%|█████████████| 4/4 [05:58<00:00, 89.60s/it]
[[1.0, 'jondurbin_bagel-dpo-34b-v0.2'], [0.8, 'NousResearch_Nous-Capybara-34B']]
07:27:40 - Layer 57/60 - CHANGED - -0.00152 > -0.00171 - 12.5%
----
Optimizing Layer 58/60 (slerp): 100%|█████████████| 4/4 [06:03<00:00, 90.92s/it]
[[1.0, 'jondurbin_bagel-dpo-34b-v0.2'], [0.6, 'NousResearch_Nous-Capybara-34B']]
07:35:25 - Layer 58/60 - CHANGED - -0.00171 > -0.00186 - 8.8%
----
Optimizing Layer 59/60 (slerp): 100%|█████████████| 4/4 [05:25<00:00, 81.34s/it]
[[1.0, 'jondurbin_bagel-dpo-34b-v0.2']]
07:42:28 - Layer 59/60 - RETAINED - -0.00186
----
Optimizing Layer 60/60 (slerp): 100%|█████████████| 4/4 [05:45<00:00, 86.41s/it]
[[1.0, 'jondurbin_bagel-dpo-34b-v0.2']]
07:49:59 - Layer 60/60 - RETAINED - -0.00186
----
Optimizing Header: 100%|██████████████████████████| 4/4 [06:22<00:00, 95.55s/it]
[[1.0, 'jondurbin_bagel-dpo-34b-v0.2'], [0.6, 'NousResearch_Nous-Capybara-34B']]
07:57:33 - Header - CHANGED - -0.00186 > -0.00190 - 2.5%
-----------------------------------------------------------------------------------------------------
| Type | Phrase | Context | Raw Prob* | Used Prob** | Change |
-----------------------------------------------------------------------------------------------------
| BAD | anticipation | Her body quivers with | 0.00000% | 0.00% | +0.00% |
| BAD | anticipation | The atmosphere is thic.. | 0.00000% | 0.00% | +0.00% |
| BAD | unwavering | Filled with an | 0.00000% | 0.00% | +0.00% |
| BAD | determination | Her eyes were filled w.. | 0.00000% | 0.00% | -0.00% |
| BAD | determination | Her stubbornness only .. | 0.00000% | 0.00% | +0.00% |
| BAD | whisper | Her voice barely above.. | 0.00000% | 0.00% | +0.00% |
| BAD | spine | shivers down her | 0.00000% | 0.00% | -0.00% |
| BAD | sends shivers | The thrill of the act | 0.00000% | 0.00% | +0.00% |
| BAD | ministrations | She moans and twitches.. | 0.00004% | 0.00% | -0.00% |
| BAD | legs | wraps her | 0.00000% | 0.00% | -0.00% |
| BAD | imposing figure | He had an | 0.00000% | 0.00% | -0.00% |
| BAD | shared challenges | Their bond strengthene.. | 0.00001% | 0.00% | +0.00% |
| BAD | bond | forged a | 0.00007% | 0.00% | -0.00% |
| BAD | bond | an unspoken | 0.00010% | 0.00% | +0.00% |
| BAD | enhance our expe.. | I'm excited to see how | 0.00000% | 0.00% | +0.00% |
| BAD | sense of vulnera.. | create a | 0.00000% | 0.00% | -0.00% |
| BAD | dimensions of in.. | explore new | 0.00000% | 0.00% | +0.00% |
| BAD | deepening our co.. | while | 0.00000% | 0.00% | -0.00% |
| BAD | shared experiences | through | 0.00000% | 0.00% | -0.00% |
| BAD | societal expecta.. | that transcend | 0.00000% | 0.00% | -0.00% |
| BAD | conventional bou.. | that defy | 0.00000% | 0.00% | +0.00% |
| BAD | conventional bou.. | and defy | 0.00000% | 0.00% | +0.00% |
| BAD | open communication | an environment | 0.00000% | 0.00% | -0.00% |
| BAD | emotional vulner.. | an environment | 0.00000% | 0.00% | -0.00% |
| BAD | heightens our co.. | touch and the anticipa.. | 0.00000% | 0.00% | -0.00% |
| BAD | sensations you'r.. | I'm enjoying | 0.00000% | 0.00% | +0.00% |
| BAD | is truly arousing | attention to detail | 0.00000% | 0.00% | +0.00% |
| BAD | is truly arousing | way you explore my body | 0.00000% | 0.00% | +0.00% |
| BAD | challenge presen.. | my resolve unwavering .. | 0.00000% | 0.00% | +0.00% |
| BAD | humble vessel | surrendering to the ex.. | 0.00000% | 0.00% | +0.00% |
| BAD | bond | cherishing the unique | 0.00017% | 0.00% | +0.00% |
| BAD | bond | special | 0.00011% | 0.00% | -0.00% |
| BAD | grows stronger w.. | bond | 0.00000% | 0.00% | -0.00% |
| BAD | that cannot be b.. | bond | 0.00000% | 0.00% | +0.00% |
| BAD | becomes unbreaka.. | bond | 0.00000% | 0.00% | -0.00% |
| BAD | grew stronger wi.. | bond | 0.00000% | 0.00% | -0.00% |
| GOOD | The apple is in .. | Question: If I'm in th.. | 0.19188% | 0.19% | +0.19% |
------------------------------------------------------------------------------------------------------
| Totals | 0.19% | 0.20% | 0.19% |
------------------------------------------------------------------------------------------------------
* = Unweighted, raw probability - ** = Probability after weight adjustments
-------- MERGE COMPOSITION ---------
jondurbin_bagel-dpo-34b-v0.2: 0.70
NousResearch_Nous-Capybara-34B: 0.30
------------------------------------
07:59:18 - Loading model (../NousResearch_Nous-Hermes-2-Yi-34B)...
Loading checkpoint shards: 100%|████████████████| 15/15 [00:33<00:00, 2.22s/it]
08:00:31 - Model loaded. Dtype: torch.float16
------------------------------------
Optimizing Layer 1/60 (slerp): 100%|██████████████| 4/4 [03:32<00:00, 53.01s/it]
[[1.0, 'jondurbin_bagel-dpo-34b-v0.2'], [0.2, 'NousResearch_Nous-Capybara-34B'], [0.8, 'NousResearch_Nous-Hermes-2-Yi-34B']]
08:05:31 - Layer 1/60 - CHANGED - -0.00186 > -0.00230 - 23.5%
----
Optimizing Layer 2/60 (slerp): 100%|██████████████| 4/4 [03:40<00:00, 55.00s/it]
[[1.0, 'jondurbin_bagel-dpo-34b-v0.2'], [0.4, 'NousResearch_Nous-Capybara-34B'], [0.8, 'NousResearch_Nous-Hermes-2-Yi-34B']]
08:10:21 - Layer 2/60 - CHANGED - -0.00230 > -0.00266 - 15.9%
----
Optimizing Layer 3/60 (slerp): 100%|██████████████| 4/4 [04:33<00:00, 68.26s/it]
[[1.0, 'jondurbin_bagel-dpo-34b-v0.2']]
08:16:22 - Layer 3/60 - RETAINED - -0.00266
----
Optimizing Layer 4/60 (slerp): 100%|██████████████| 4/4 [05:06<00:00, 76.71s/it]
[[1.0, 'jondurbin_bagel-dpo-34b-v0.2'], [0.8, 'NousResearch_Nous-Hermes-2-Yi-34B']]
08:23:09 - Layer 4/60 - CHANGED - -0.00266 > -0.00294 - 10.5%
----
Optimizing Layer 5/60 (slerp): 100%|██████████████| 4/4 [05:47<00:00, 86.79s/it]
[[1.0, 'jondurbin_bagel-dpo-34b-v0.2']]
08:30:35 - Layer 5/60 - RETAINED - -0.00294
----
Optimizing Layer 6/60 (slerp): 100%|██████████████| 4/4 [05:25<00:00, 81.41s/it]
[[1.0, 'jondurbin_bagel-dpo-34b-v0.2']]
08:37:52 - Layer 6/60 - RETAINED - -0.00294
----
Optimizing Layer 7/60 (slerp): 100%|██████████████| 4/4 [05:44<00:00, 86.12s/it]
[[1.0, 'jondurbin_bagel-dpo-34b-v0.2']]
08:45:26 - Layer 7/60 - RETAINED - -0.00294
----
Optimizing Layer 8/60 (slerp): 100%|██████████████| 4/4 [05:36<00:00, 84.21s/it]
[[1.0, 'jondurbin_bagel-dpo-34b-v0.2']]
08:52:56 - Layer 8/60 - RETAINED - -0.00294
----
Optimizing Layer 9/60 (slerp): 100%|██████████████| 4/4 [05:51<00:00, 87.81s/it]
[[1.0, 'jondurbin_bagel-dpo-34b-v0.2'], [0.4, 'NousResearch_Nous-Capybara-34B'], [0.6, 'NousResearch_Nous-Hermes-2-Yi-34B']]
09:00:30 - Layer 9/60 - CHANGED - -0.00294 > -0.00297 - 1.2%
----
Optimizing Layer 10/60 (slerp): 100%|█████████████| 4/4 [06:03<00:00, 90.97s/it]
[[1.0, 'jondurbin_bagel-dpo-34b-v0.2']]
09:08:29 - Layer 10/60 - RETAINED - -0.00297
----
Optimizing Layer 11/60 (slerp): 100%|█████████████| 4/4 [05:19<00:00, 79.95s/it]
[[1.0, 'jondurbin_bagel-dpo-34b-v0.2'], [0.8, 'NousResearch_Nous-Capybara-34B'], [0.8, 'NousResearch_Nous-Hermes-2-Yi-34B']]
09:15:40 - Layer 11/60 - CHANGED - -0.00297 > -0.00334 - 12.2%
----
Optimizing Layer 12/60 (slerp): 100%|█████████████| 4/4 [05:47<00:00, 86.85s/it]
[[1.0, 'jondurbin_bagel-dpo-34b-v0.2'], [0.8, 'NousResearch_Nous-Capybara-34B']]
09:23:46 - Layer 12/60 - RETAINED - -0.00334
----
Optimizing Layer 13/60 (slerp): 100%|█████████████| 4/4 [05:05<00:00, 76.33s/it]
[[1.0, 'jondurbin_bagel-dpo-34b-v0.2'], [0.4, 'NousResearch_Nous-Capybara-34B']]
09:30:37 - Layer 13/60 - RETAINED - -0.00334
----
Optimizing Layer 14/60 (slerp): 100%|█████████████| 4/4 [04:47<00:00, 71.79s/it]
[[1.0, 'jondurbin_bagel-dpo-34b-v0.2'], [0.4, 'NousResearch_Nous-Capybara-34B'], [0.2, 'NousResearch_Nous-Hermes-2-Yi-34B']]
09:37:17 - Layer 14/60 - CHANGED - -0.00334 > -0.00336 - 0.8%
----
Optimizing Layer 15/60 (slerp): 100%|█████████████| 4/4 [04:05<00:00, 61.32s/it]
[[1.0, 'jondurbin_bagel-dpo-34b-v0.2']]
09:42:46 - Layer 15/60 - RETAINED - -0.00336
----
Optimizing Layer 16/60 (slerp): 100%|█████████████| 4/4 [04:16<00:00, 64.24s/it]
[[1.0, 'jondurbin_bagel-dpo-34b-v0.2']]
09:48:30 - Layer 16/60 - RETAINED - -0.00336
----
Optimizing Layer 17/60 (slerp): 100%|█████████████| 4/4 [04:31<00:00, 67.78s/it]
[[1.0, 'jondurbin_bagel-dpo-34b-v0.2'], [0.6, 'NousResearch_Nous-Hermes-2-Yi-34B']]
09:54:37 - Layer 17/60 - CHANGED - -0.00336 > -0.00361 - 7.3%
----
Optimizing Layer 18/60 (slerp): 100%|█████████████| 4/4 [04:35<00:00, 68.88s/it]
[[1.0, 'jondurbin_bagel-dpo-34b-v0.2']]
10:00:44 - Layer 18/60 - RETAINED - -0.00361
----
Optimizing Layer 19/60 (slerp): 100%|█████████████| 4/4 [05:48<00:00, 87.17s/it]
[[1.0, 'jondurbin_bagel-dpo-34b-v0.2']]
10:08:21 - Layer 19/60 - RETAINED - -0.00361
----
Optimizing Layer 20/60 (slerp): 100%|█████████████| 4/4 [05:12<00:00, 78.07s/it]
[[1.0, 'jondurbin_bagel-dpo-34b-v0.2'], [0.8, 'NousResearch_Nous-Capybara-34B']]
10:15:11 - Layer 20/60 - RETAINED - -0.00361
----
Optimizing Layer 21/60 (slerp): 100%|█████████████| 4/4 [04:18<00:00, 64.71s/it]
[[1.0, 'jondurbin_bagel-dpo-34b-v0.2'], [0.8, 'NousResearch_Nous-Capybara-34B'], [0.8, 'NousResearch_Nous-Hermes-2-Yi-34B']]
10:20:54 - Layer 21/60 - CHANGED - -0.00361 > -0.00376 - 4.3%
----
Optimizing Layer 22/60 (slerp): 100%|█████████████| 4/4 [03:46<00:00, 56.73s/it]
[[1.0, 'jondurbin_bagel-dpo-34b-v0.2'], [0.8, 'NousResearch_Nous-Capybara-34B'], [0.8, 'NousResearch_Nous-Hermes-2-Yi-34B']]
10:26:01 - Layer 22/60 - CHANGED - -0.00376 > -0.00466 - 24.0%
----
Optimizing Layer 23/60 (slerp): 100%|█████████████| 4/4 [04:01<00:00, 60.46s/it]
[[1.0, 'jondurbin_bagel-dpo-34b-v0.2'], [0.8, 'NousResearch_Nous-Capybara-34B'], [0.4, 'NousResearch_Nous-Hermes-2-Yi-34B']]
10:31:21 - Layer 23/60 - CHANGED - -0.00466 > -0.00616 - 32.1%
----
Optimizing Layer 24/60 (slerp): 100%|█████████████| 4/4 [04:06<00:00, 61.57s/it]
[[1.0, 'jondurbin_bagel-dpo-34b-v0.2'], [0.8, 'NousResearch_Nous-Capybara-34B'], [0.8, 'NousResearch_Nous-Hermes-2-Yi-34B']]
10:36:43 - Layer 24/60 - CHANGED - -0.00616 > -0.00743 - 20.6%
----
Optimizing Layer 25/60 (slerp): 100%|█████████████| 4/4 [04:09<00:00, 62.32s/it]
[[1.0, 'jondurbin_bagel-dpo-34b-v0.2'], [0.8, 'NousResearch_Nous-Capybara-34B']]
10:42:19 - Layer 25/60 - RETAINED - -0.00743
----
Optimizing Layer 26/60 (slerp): 100%|█████████████| 4/4 [04:27<00:00, 66.78s/it]
[[1.0, 'jondurbin_bagel-dpo-34b-v0.2'], [0.4, 'NousResearch_Nous-Capybara-34B'], [0.4, 'NousResearch_Nous-Hermes-2-Yi-34B']]
10:48:17 - Layer 26/60 - CHANGED - -0.00743 > -0.00745 - 0.3%
----
Optimizing Layer 27/60 (slerp): 100%|█████████████| 4/4 [05:11<00:00, 77.78s/it]
[[1.0, 'jondurbin_bagel-dpo-34b-v0.2']]
10:55:12 - Layer 27/60 - RETAINED - -0.00745
----
Optimizing Layer 28/60 (slerp): 100%|█████████████| 4/4 [05:31<00:00, 82.92s/it]
[[1.0, 'jondurbin_bagel-dpo-34b-v0.2'], [0.4, 'NousResearch_Nous-Capybara-34B'], [0.8, 'NousResearch_Nous-Hermes-2-Yi-34B']]
11:02:26 - Layer 28/60 - CHANGED - -0.00745 > -0.00789 - 5.9%
----
Optimizing Layer 29/60 (slerp): 100%|█████████████| 4/4 [05:10<00:00, 77.75s/it]
[[1.0, 'jondurbin_bagel-dpo-34b-v0.2'], [0.8, 'NousResearch_Nous-Capybara-34B'], [0.8, 'NousResearch_Nous-Hermes-2-Yi-34B']]
11:09:12 - Layer 29/60 - CHANGED - -0.00789 > -0.00824 - 4.5%
----
Optimizing Layer 30/60 (slerp): 100%|█████████████| 4/4 [05:35<00:00, 83.82s/it]
[[1.0, 'jondurbin_bagel-dpo-34b-v0.2'], [0.2, 'NousResearch_Nous-Capybara-34B'], [0.8, 'NousResearch_Nous-Hermes-2-Yi-34B']]
11:16:32 - Layer 30/60 - CHANGED - -0.00824 > -0.00980 - 18.9%
----
Optimizing Layer 31/60 (slerp): 100%|█████████████| 4/4 [06:09<00:00, 92.45s/it]
[[1.0, 'jondurbin_bagel-dpo-34b-v0.2'], [0.8, 'NousResearch_Nous-Capybara-34B'], [0.8, 'NousResearch_Nous-Hermes-2-Yi-34B']]
11:24:35 - Layer 31/60 - CHANGED - -0.00980 > -0.01486 - 51.6%
----
Optimizing Layer 32/60 (slerp): 100%|█████████████| 4/4 [05:35<00:00, 83.93s/it]
[[1.0, 'jondurbin_bagel-dpo-34b-v0.2'], [0.6, 'NousResearch_Nous-Hermes-2-Yi-34B']]
11:32:09 - Layer 32/60 - CHANGED - -0.01486 > -0.01743 - 17.3%
----
Optimizing Layer 33/60 (slerp): 100%|█████████████| 4/4 [05:40<00:00, 85.07s/it]
[[1.0, 'jondurbin_bagel-dpo-34b-v0.2']]
11:39:27 - Layer 33/60 - RETAINED - -0.01743
----
Optimizing Layer 34/60 (slerp): 100%|█████████████| 4/4 [05:28<00:00, 82.20s/it]
[[1.0, 'jondurbin_bagel-dpo-34b-v0.2'], [0.8, 'NousResearch_Nous-Hermes-2-Yi-34B']]
11:46:40 - Layer 34/60 - CHANGED - -0.01743 > -0.02148 - 23.2%
----
Optimizing Layer 35/60 (slerp): 100%|█████████████| 4/4 [06:17<00:00, 94.36s/it]
[[1.0, 'jondurbin_bagel-dpo-34b-v0.2']]
11:54:42 - Layer 35/60 - RETAINED - -0.02148
----
Optimizing Layer 36/60 (slerp): 100%|█████████████| 4/4 [05:46<00:00, 86.54s/it]
[[1.0, 'jondurbin_bagel-dpo-34b-v0.2'], [0.8, 'NousResearch_Nous-Capybara-34B']]
12:02:23 - Layer 36/60 - RETAINED - -0.02148
----
Optimizing Layer 37/60 (slerp): 100%|█████████████| 4/4 [04:44<00:00, 71.19s/it]
[[1.0, 'jondurbin_bagel-dpo-34b-v0.2'], [0.8, 'NousResearch_Nous-Capybara-34B'], [0.8, 'NousResearch_Nous-Hermes-2-Yi-34B']]
12:08:46 - Layer 37/60 - CHANGED - -0.02148 > -0.02760 - 28.5%
----
Optimizing Layer 38/60 (slerp): 100%|█████████████| 4/4 [03:58<00:00, 59.73s/it]
[[1.0, 'jondurbin_bagel-dpo-34b-v0.2'], [0.4, 'NousResearch_Nous-Hermes-2-Yi-34B']]
12:14:11 - Layer 38/60 - CHANGED - -0.02760 > -0.02789 - 1.0%
----
Optimizing Layer 39/60 (slerp): 100%|█████████████| 4/4 [04:00<00:00, 60.16s/it]
[[1.0, 'jondurbin_bagel-dpo-34b-v0.2'], [0.8, 'NousResearch_Nous-Capybara-34B']]
12:19:28 - Layer 39/60 - RETAINED - -0.02789
----
Optimizing Layer 40/60 (slerp): 100%|█████████████| 4/4 [03:57<00:00, 59.45s/it]
[[1.0, 'jondurbin_bagel-dpo-34b-v0.2']]
12:24:49 - Layer 40/60 - RETAINED - -0.02789
----
Optimizing Layer 41/60 (slerp): 100%|█████████████| 4/4 [04:01<00:00, 60.34s/it]
[[1.0, 'jondurbin_bagel-dpo-34b-v0.2']]
12:30:08 - Layer 41/60 - RETAINED - -0.02789
----
Optimizing Layer 42/60 (slerp): 100%|█████████████| 4/4 [04:01<00:00, 60.29s/it]
[[1.0, 'jondurbin_bagel-dpo-34b-v0.2']]
12:35:23 - Layer 42/60 - RETAINED - -0.02789
----
Optimizing Layer 43/60 (slerp): 100%|█████████████| 4/4 [04:18<00:00, 64.70s/it]
[[1.0, 'jondurbin_bagel-dpo-34b-v0.2']]
12:41:09 - Layer 43/60 - RETAINED - -0.02789
----
Optimizing Layer 44/60 (slerp): 100%|█████████████| 4/4 [04:44<00:00, 71.20s/it]
[[1.0, 'jondurbin_bagel-dpo-34b-v0.2']]
12:47:23 - Layer 44/60 - RETAINED - -0.02789
----
Optimizing Layer 45/60 (slerp): 100%|█████████████| 4/4 [03:42<00:00, 55.71s/it]
[[1.0, 'jondurbin_bagel-dpo-34b-v0.2']]
12:52:31 - Layer 45/60 - RETAINED - -0.02789
----
Optimizing Layer 46/60 (slerp): 100%|█████████████| 4/4 [03:59<00:00, 59.77s/it]
[[1.0, 'jondurbin_bagel-dpo-34b-v0.2']]
12:57:52 - Layer 46/60 - RETAINED - -0.02789
----
Optimizing Layer 47/60 (slerp): 100%|█████████████| 4/4 [04:03<00:00, 60.98s/it]
[[1.0, 'jondurbin_bagel-dpo-34b-v0.2']]
13:03:16 - Layer 47/60 - RETAINED - -0.02789
----
Optimizing Layer 48/60 (slerp): 100%|█████████████| 4/4 [03:53<00:00, 58.40s/it]
[[1.0, 'jondurbin_bagel-dpo-34b-v0.2'], [0.2, 'NousResearch_Nous-Hermes-2-Yi-34B']]
13:08:28 - Layer 48/60 - CHANGED - -0.02789 > -0.02789 - 0.0%
----
Optimizing Layer 49/60 (slerp): 100%|█████████████| 4/4 [03:57<00:00, 59.32s/it]
[[1.0, 'jondurbin_bagel-dpo-34b-v0.2'], [0.4, 'NousResearch_Nous-Capybara-34B'], [0.8, 'NousResearch_Nous-Hermes-2-Yi-34B']]
13:13:43 - Layer 49/60 - CHANGED - -0.02789 > -0.02922 - 4.8%
----
Optimizing Layer 50/60 (slerp): 100%|█████████████| 4/4 [04:03<00:00, 60.93s/it]
[[1.0, 'jondurbin_bagel-dpo-34b-v0.2'], [0.6, 'NousResearch_Nous-Capybara-34B'], [0.8, 'NousResearch_Nous-Hermes-2-Yi-34B']]
13:19:09 - Layer 50/60 - CHANGED - -0.02922 > -0.03467 - 18.6%
----
Optimizing Layer 51/60 (slerp): 100%|█████████████| 4/4 [04:06<00:00, 61.73s/it]
[[1.0, 'jondurbin_bagel-dpo-34b-v0.2']]
13:24:39 - Layer 51/60 - RETAINED - -0.03467
----
Optimizing Layer 52/60 (slerp): 100%|█████████████| 4/4 [04:02<00:00, 60.70s/it]
[[1.0, 'jondurbin_bagel-dpo-34b-v0.2'], [0.8, 'NousResearch_Nous-Capybara-34B'], [0.8, 'NousResearch_Nous-Hermes-2-Yi-34B']]
13:29:58 - Layer 52/60 - CHANGED - -0.03467 > -0.03931 - 13.4%
----
Optimizing Layer 53/60 (slerp): 100%|█████████████| 4/4 [04:00<00:00, 60.06s/it]
[[1.0, 'jondurbin_bagel-dpo-34b-v0.2'], [0.2, 'NousResearch_Nous-Capybara-34B'], [0.4, 'NousResearch_Nous-Hermes-2-Yi-34B']]
13:35:19 - Layer 53/60 - CHANGED - -0.03931 > -0.04040 - 2.8%
----
Optimizing Layer 54/60 (slerp): 100%|█████████████| 4/4 [04:30<00:00, 67.51s/it]
[[1.0, 'jondurbin_bagel-dpo-34b-v0.2'], [0.4, 'NousResearch_Nous-Capybara-34B'], [0.8, 'NousResearch_Nous-Hermes-2-Yi-34B']]
13:41:14 - Layer 54/60 - CHANGED - -0.04040 > -0.04498 - 11.3%
----
Optimizing Layer 55/60 (slerp): 100%|█████████████| 4/4 [04:50<00:00, 72.65s/it]
[[1.0, 'jondurbin_bagel-dpo-34b-v0.2'], [0.8, 'NousResearch_Nous-Hermes-2-Yi-34B']]
13:47:49 - Layer 55/60 - CHANGED - -0.04498 > -0.04736 - 5.3%
----
Optimizing Layer 56/60 (slerp): 100%|█████████████| 4/4 [05:28<00:00, 82.16s/it]
[[1.0, 'jondurbin_bagel-dpo-34b-v0.2'], [0.8, 'NousResearch_Nous-Capybara-34B']]
13:55:09 - Layer 56/60 - RETAINED - -0.04736
----
Optimizing Layer 57/60 (slerp): 100%|█████████████| 4/4 [05:30<00:00, 82.57s/it]
[[1.0, 'jondurbin_bagel-dpo-34b-v0.2'], [0.8, 'NousResearch_Nous-Capybara-34B']]
14:02:30 - Layer 57/60 - RETAINED - -0.04736
----
Optimizing Layer 58/60 (slerp): 100%|█████████████| 4/4 [06:22<00:00, 95.56s/it]
[[1.0, 'jondurbin_bagel-dpo-34b-v0.2'], [0.6, 'NousResearch_Nous-Capybara-34B']]
14:11:07 - Layer 58/60 - RETAINED - -0.04736
----
Optimizing Layer 59/60 (slerp): 100%|█████████████| 4/4 [05:52<00:00, 88.03s/it]
[[1.0, 'jondurbin_bagel-dpo-34b-v0.2'], [0.8, 'NousResearch_Nous-Hermes-2-Yi-34B']]
14:19:17 - Layer 59/60 - CHANGED - -0.04736 > -0.05244 - 10.7%
----
Optimizing Layer 60/60 (slerp): 100%|█████████████| 4/4 [04:47<00:00, 71.86s/it]
[[1.0, 'jondurbin_bagel-dpo-34b-v0.2']]
14:25:42 - Layer 60/60 - RETAINED - -0.05244
----
Optimizing Header: 100%|██████████████████████████| 4/4 [03:37<00:00, 54.33s/it]
[[1.0, 'jondurbin_bagel-dpo-34b-v0.2'], [0.6, 'NousResearch_Nous-Capybara-34B'], [0.8, 'NousResearch_Nous-Hermes-2-Yi-34B']]
14:30:24 - Header - CHANGED - -0.05244 > -0.06200 - 18.2%
-----------------------------------------------------------------------------------------------------
| Type | Phrase | Context | Raw Prob* | Used Prob** | Change |
-----------------------------------------------------------------------------------------------------
| BAD | anticipation | Her body quivers with | 0.00000% | 0.00% | +0.00% |
| BAD | anticipation | The atmosphere is thic.. | 0.00000% | 0.00% | +0.00% |
| BAD | unwavering | Filled with an | 0.00000% | 0.00% | +0.00% |
| BAD | determination | Her eyes were filled w.. | 0.00000% | 0.00% | -0.00% |
| BAD | determination | Her stubbornness only .. | 0.00000% | 0.00% | +0.00% |
| BAD | whisper | Her voice barely above.. | 0.00000% | 0.00% | +0.00% |
| BAD | spine | shivers down her | 0.00000% | 0.00% | +0.00% |
| BAD | sends shivers | The thrill of the act | 0.00000% | 0.00% | +0.00% |
| BAD | ministrations | She moans and twitches.. | 0.00003% | 0.00% | -0.00% |
| BAD | legs | wraps her | 0.00000% | 0.00% | -0.00% |
| BAD | imposing figure | He had an | 0.00000% | 0.00% | -0.00% |
| BAD | shared challenges | Their bond strengthene.. | 0.00001% | 0.00% | +0.00% |
| BAD | bond | forged a | 0.00004% | 0.00% | -0.00% |
| BAD | bond | an unspoken | 0.00010% | 0.00% | +0.00% |
| BAD | enhance our expe.. | I'm excited to see how | 0.00000% | 0.00% | +0.00% |
| BAD | sense of vulnera.. | create a | 0.00000% | 0.00% | -0.00% |
| BAD | dimensions of in.. | explore new | 0.00000% | 0.00% | +0.00% |
| BAD | deepening our co.. | while | 0.00000% | 0.00% | -0.00% |
| BAD | shared experiences | through | 0.00001% | 0.00% | +0.00% |
| BAD | societal expecta.. | that transcend | 0.00000% | 0.00% | -0.00% |
| BAD | conventional bou.. | that defy | 0.00000% | 0.00% | +0.00% |
| BAD | conventional bou.. | and defy | 0.00000% | 0.00% | +0.00% |
| BAD | open communication | an environment | 0.00000% | 0.00% | +0.00% |
| BAD | emotional vulner.. | an environment | 0.00000% | 0.00% | +0.00% |
| BAD | heightens our co.. | touch and the anticipa.. | 0.00000% | 0.00% | -0.00% |
| BAD | sensations you'r.. | I'm enjoying | 0.00000% | 0.00% | +0.00% |
| BAD | is truly arousing | attention to detail | 0.00000% | 0.00% | +0.00% |
| BAD | is truly arousing | way you explore my body | 0.00000% | 0.00% | -0.00% |
| BAD | challenge presen.. | my resolve unwavering .. | 0.00000% | 0.00% | +0.00% |
| BAD | humble vessel | surrendering to the ex.. | 0.00000% | 0.00% | +0.00% |
| BAD | bond | cherishing the unique | 0.00019% | 0.00% | +0.00% |
| BAD | bond | special | 0.00023% | 0.00% | -0.00% |
| BAD | grows stronger w.. | bond | 0.00000% | 0.00% | -0.00% |
| BAD | that cannot be b.. | bond | 0.00000% | 0.00% | -0.00% |
| BAD | becomes unbreaka.. | bond | 0.00000% | 0.00% | +0.00% |
| BAD | grew stronger wi.. | bond | 0.00000% | 0.00% | +0.00% |
| GOOD | The apple is in .. | Question: If I'm in th.. | 6.12871% | 6.13% | +6.13% |
------------------------------------------------------------------------------------------------------
| Totals | 6.13% | 6.14% | 6.13% |
------------------------------------------------------------------------------------------------------
* = Unweighted, raw probability - ** = Probability after weight adjustments
-------- MERGE COMPOSITION ---------
jondurbin_bagel-dpo-34b-v0.2: 0.51
NousResearch_Nous-Hermes-2-Yi-34B: 0.32
NousResearch_Nous-Capybara-34B: 0.16
------------------------------------
14:31:32 - Loading model (../SUSTech_SUS-Chat-34B)...
Loading checkpoint shards: 100%|██████████████████| 7/7 [01:14<00:00, 10.68s/it]
14:33:15 - Model loaded. Dtype: torch.float16
------------------------------------
Optimizing Layer 1/60 (slerp): 100%|██████████████| 4/4 [02:55<00:00, 43.98s/it]
[[1.0, 'jondurbin_bagel-dpo-34b-v0.2'], [0.2, 'NousResearch_Nous-Capybara-34B'], [0.8, 'NousResearch_Nous-Hermes-2-Yi-34B'], [0.2, 'SUSTech_SUS-Chat-34B']]
14:37:13 - Layer 1/60 - CHANGED - -0.06121 > -0.06153 - 0.5%
----
Optimizing Layer 2/60 (slerp): 100%|██████████████| 4/4 [02:57<00:00, 44.28s/it]
[[1.0, 'jondurbin_bagel-dpo-34b-v0.2'], [0.4, 'NousResearch_Nous-Capybara-34B'], [0.8, 'NousResearch_Nous-Hermes-2-Yi-34B'], [0.6, 'SUSTech_SUS-Chat-34B']]
14:41:08 - Layer 2/60 - CHANGED - -0.06153 > -0.06434 - 4.6%
----
Optimizing Layer 3/60 (slerp): 100%|██████████████| 4/4 [02:59<00:00, 44.87s/it]
[[1.0, 'jondurbin_bagel-dpo-34b-v0.2']]
14:45:04 - Layer 3/60 - RETAINED - -0.06434
----
Optimizing Layer 4/60 (slerp): 100%|██████████████| 4/4 [03:24<00:00, 51.23s/it]
[[1.0, 'jondurbin_bagel-dpo-34b-v0.2'], [0.8, 'NousResearch_Nous-Hermes-2-Yi-34B']]
14:49:33 - Layer 4/60 - RETAINED - -0.06434
----
Optimizing Layer 5/60 (slerp): 100%|██████████████| 4/4 [04:13<00:00, 63.44s/it]
[[1.0, 'jondurbin_bagel-dpo-34b-v0.2']]
14:55:17 - Layer 5/60 - RETAINED - -0.06434
----
Optimizing Layer 6/60 (slerp): 100%|██████████████| 4/4 [05:08<00:00, 77.18s/it]
[[1.0, 'jondurbin_bagel-dpo-34b-v0.2']]
15:01:58 - Layer 6/60 - RETAINED - -0.06434
----
Optimizing Layer 7/60 (slerp): 100%|██████████████| 4/4 [04:41<00:00, 70.31s/it]
[[1.0, 'jondurbin_bagel-dpo-34b-v0.2']]
15:08:01 - Layer 7/60 - RETAINED - -0.06434
----
Optimizing Layer 8/60 (slerp): 100%|██████████████| 4/4 [03:51<00:00, 57.86s/it]
[[1.0, 'jondurbin_bagel-dpo-34b-v0.2']]
15:13:10 - Layer 8/60 - RETAINED - -0.06434
----
Optimizing Layer 9/60 (slerp): 100%|██████████████| 4/4 [04:02<00:00, 60.54s/it]
[[1.0, 'jondurbin_bagel-dpo-34b-v0.2'], [0.4, 'NousResearch_Nous-Capybara-34B'], [0.6, 'NousResearch_Nous-Hermes-2-Yi-34B'], [0.4, 'SUSTech_SUS-Chat-34B']]
15:18:34 - Layer 9/60 - CHANGED - -0.06434 > -0.06464 - 0.5%
----
Optimizing Layer 10/60 (slerp): 100%|█████████████| 4/4 [03:53<00:00, 58.40s/it]
[[1.0, 'jondurbin_bagel-dpo-34b-v0.2']]
15:23:40 - Layer 10/60 - RETAINED - -0.06464
----
Optimizing Layer 11/60 (slerp): 100%|█████████████| 4/4 [03:39<00:00, 54.91s/it]
[[1.0, 'jondurbin_bagel-dpo-34b-v0.2'], [0.8, 'NousResearch_Nous-Capybara-34B'], [0.8, 'NousResearch_Nous-Hermes-2-Yi-34B']]
15:28:32 - Layer 11/60 - RETAINED - -0.06464
----
Optimizing Layer 12/60 (slerp): 100%|█████████████| 4/4 [03:40<00:00, 55.10s/it]
[[1.0, 'jondurbin_bagel-dpo-34b-v0.2'], [0.8, 'NousResearch_Nous-Capybara-34B']]
15:33:27 - Layer 12/60 - RETAINED - -0.06464
----
Optimizing Layer 13/60 (slerp): 100%|█████████████| 4/4 [03:49<00:00, 57.36s/it]
[[1.0, 'jondurbin_bagel-dpo-34b-v0.2'], [0.4, 'NousResearch_Nous-Capybara-34B'], [0.4, 'SUSTech_SUS-Chat-34B']]
15:38:35 - Layer 13/60 - CHANGED - -0.06464 > -0.06527 - 1.0%
----
Optimizing Layer 14/60 (slerp): 100%|█████████████| 4/4 [03:42<00:00, 55.74s/it]
[[1.0, 'jondurbin_bagel-dpo-34b-v0.2'], [0.4, 'NousResearch_Nous-Capybara-34B'], [0.2, 'NousResearch_Nous-Hermes-2-Yi-34B'], [0.6, 'SUSTech_SUS-Chat-34B']]
15:43:30 - Layer 14/60 - CHANGED - -0.06527 > -0.06851 - 5.0%
----
Optimizing Layer 15/60 (slerp): 100%|█████████████| 4/4 [03:44<00:00, 56.04s/it]
[[1.0, 'jondurbin_bagel-dpo-34b-v0.2']]
15:48:41 - Layer 15/60 - RETAINED - -0.06851
----
Optimizing Layer 16/60 (slerp): 100%|█████████████| 4/4 [05:23<00:00, 80.84s/it]
[[1.0, 'jondurbin_bagel-dpo-34b-v0.2']]
15:55:48 - Layer 16/60 - RETAINED - -0.06851
----
Optimizing Layer 17/60 (slerp): 100%|█████████████| 4/4 [05:31<00:00, 82.76s/it]
[[1.0, 'jondurbin_bagel-dpo-34b-v0.2'], [0.6, 'NousResearch_Nous-Hermes-2-Yi-34B']]
16:03:01 - Layer 17/60 - RETAINED - -0.06851
----
Optimizing Layer 18/60 (slerp): 100%|█████████████| 4/4 [05:34<00:00, 83.64s/it]
[[1.0, 'jondurbin_bagel-dpo-34b-v0.2']]
16:10:28 - Layer 18/60 - RETAINED - -0.06851
----
Optimizing Layer 19/60 (slerp): 100%|█████████████| 4/4 [06:17<00:00, 94.38s/it]
[[1.0, 'jondurbin_bagel-dpo-34b-v0.2']]
16:18:46 - Layer 19/60 - RETAINED - -0.06851
----
Optimizing Layer 20/60 (slerp): 100%|█████████████| 4/4 [04:52<00:00, 73.08s/it]
[[1.0, 'jondurbin_bagel-dpo-34b-v0.2'], [0.8, 'NousResearch_Nous-Capybara-34B'], [0.4, 'SUSTech_SUS-Chat-34B']]
16:25:26 - Layer 20/60 - CHANGED - -0.06851 > -0.06892 - 0.6%
----
Optimizing Layer 21/60 (slerp): 100%|█████████████| 4/4 [05:08<00:00, 77.11s/it]
[[1.0, 'jondurbin_bagel-dpo-34b-v0.2'], [0.8, 'NousResearch_Nous-Capybara-34B'], [0.8, 'NousResearch_Nous-Hermes-2-Yi-34B']]
16:32:37 - Layer 21/60 - RETAINED - -0.06892
----
Optimizing Layer 22/60 (slerp): 100%|█████████████| 4/4 [04:54<00:00, 73.54s/it]
[[1.0, 'jondurbin_bagel-dpo-34b-v0.2'], [0.8, 'NousResearch_Nous-Capybara-34B'], [0.8, 'NousResearch_Nous-Hermes-2-Yi-34B']]
16:39:05 - Layer 22/60 - RETAINED - -0.06892
----
Optimizing Layer 23/60 (slerp): 100%|█████████████| 4/4 [04:53<00:00, 73.34s/it]
[[1.0, 'jondurbin_bagel-dpo-34b-v0.2'], [0.8, 'NousResearch_Nous-Capybara-34B'], [0.4, 'NousResearch_Nous-Hermes-2-Yi-34B']]
16:45:29 - Layer 23/60 - RETAINED - -0.06892
----
Optimizing Layer 24/60 (slerp): 100%|█████████████| 4/4 [04:53<00:00, 73.38s/it]
[[1.0, 'jondurbin_bagel-dpo-34b-v0.2'], [0.8, 'NousResearch_Nous-Capybara-34B'], [0.8, 'NousResearch_Nous-Hermes-2-Yi-34B']]
16:51:58 - Layer 24/60 - RETAINED - -0.06892
----
Optimizing Layer 25/60 (slerp): 100%|█████████████| 4/4 [04:55<00:00, 73.86s/it]
[[1.0, 'jondurbin_bagel-dpo-34b-v0.2'], [0.8, 'NousResearch_Nous-Capybara-34B'], [0.4, 'SUSTech_SUS-Chat-34B']]
16:58:30 - Layer 25/60 - CHANGED - -0.06892 > -0.07074 - 2.6%
----
Optimizing Layer 26/60 (slerp): 100%|█████████████| 4/4 [04:11<00:00, 62.83s/it]
[[1.0, 'jondurbin_bagel-dpo-34b-v0.2'], [0.4, 'NousResearch_Nous-Capybara-34B'], [0.4, 'NousResearch_Nous-Hermes-2-Yi-34B']]
17:04:08 - Layer 26/60 - RETAINED - -0.07074
----
Optimizing Layer 27/60 (slerp): 100%|█████████████| 4/4 [04:11<00:00, 62.75s/it]
[[1.0, 'jondurbin_bagel-dpo-34b-v0.2']]
17:09:50 - Layer 27/60 - RETAINED - -0.07074
----
Optimizing Layer 28/60 (slerp): 100%|█████████████| 4/4 [04:05<00:00, 61.40s/it]
[[1.0, 'jondurbin_bagel-dpo-34b-v0.2'], [0.4, 'NousResearch_Nous-Capybara-34B'], [0.8, 'NousResearch_Nous-Hermes-2-Yi-34B']]
17:15:21 - Layer 28/60 - RETAINED - -0.07074
----
Optimizing Layer 29/60 (slerp): 100%|█████████████| 4/4 [05:07<00:00, 76.83s/it]
[[1.0, 'jondurbin_bagel-dpo-34b-v0.2'], [0.8, 'NousResearch_Nous-Capybara-34B'], [0.8, 'NousResearch_Nous-Hermes-2-Yi-34B']]
17:21:57 - Layer 29/60 - RETAINED - -0.07074
----
Optimizing Layer 30/60 (slerp): 100%|█████████████| 4/4 [04:06<00:00, 61.63s/it]
[[1.0, 'jondurbin_bagel-dpo-34b-v0.2'], [0.2, 'NousResearch_Nous-Capybara-34B'], [0.8, 'NousResearch_Nous-Hermes-2-Yi-34B']]
17:27:34 - Layer 30/60 - RETAINED - -0.07074
----
Optimizing Layer 31/60 (slerp): 100%|█████████████| 4/4 [04:21<00:00, 65.25s/it]
[[1.0, 'jondurbin_bagel-dpo-34b-v0.2'], [0.8, 'NousResearch_Nous-Capybara-34B'], [0.8, 'NousResearch_Nous-Hermes-2-Yi-34B']]
17:33:24 - Layer 31/60 - RETAINED - -0.07074
----
Optimizing Layer 32/60 (slerp): 100%|█████████████| 4/4 [04:36<00:00, 69.13s/it]
[[1.0, 'jondurbin_bagel-dpo-34b-v0.2'], [0.6, 'NousResearch_Nous-Hermes-2-Yi-34B']]
17:39:20 - Layer 32/60 - RETAINED - -0.07074
----
Optimizing Layer 33/60 (slerp): 100%|█████████████| 4/4 [04:52<00:00, 73.01s/it]
[[1.0, 'jondurbin_bagel-dpo-34b-v0.2']]
17:45:42 - Layer 33/60 - RETAINED - -0.07074
----
Optimizing Layer 34/60 (slerp): 100%|█████████████| 4/4 [05:09<00:00, 77.30s/it]
[[1.0, 'jondurbin_bagel-dpo-34b-v0.2'], [0.8, 'NousResearch_Nous-Hermes-2-Yi-34B']]
17:52:34 - Layer 34/60 - RETAINED - -0.07074
----
Optimizing Layer 35/60 (slerp): 100%|█████████████| 4/4 [05:09<00:00, 77.29s/it]
[[1.0, 'jondurbin_bagel-dpo-34b-v0.2']]
17:59:16 - Layer 35/60 - RETAINED - -0.07074
----
Optimizing Layer 36/60 (slerp): 100%|█████████████| 4/4 [05:19<00:00, 79.91s/it]
[[1.0, 'jondurbin_bagel-dpo-34b-v0.2'], [0.8, 'NousResearch_Nous-Capybara-34B']]
18:06:13 - Layer 36/60 - RETAINED - -0.07074
----
Optimizing Layer 37/60 (slerp): 100%|█████████████| 4/4 [05:40<00:00, 85.08s/it]
[[1.0, 'jondurbin_bagel-dpo-34b-v0.2'], [0.8, 'NousResearch_Nous-Capybara-34B'], [0.8, 'NousResearch_Nous-Hermes-2-Yi-34B'], [0.8, 'SUSTech_SUS-Chat-34B']]
18:13:35 - Layer 37/60 - CHANGED - -0.07074 > -0.07127 - 0.8%
----
Optimizing Layer 38/60 (slerp): 100%|█████████████| 4/4 [04:50<00:00, 72.69s/it]
[[1.0, 'jondurbin_bagel-dpo-34b-v0.2'], [0.4, 'NousResearch_Nous-Hermes-2-Yi-34B']]
18:20:03 - Layer 38/60 - RETAINED - -0.07127
----
Optimizing Layer 39/60 (slerp): 100%|█████████████| 4/4 [05:23<00:00, 80.96s/it]
[[1.0, 'jondurbin_bagel-dpo-34b-v0.2'], [0.8, 'NousResearch_Nous-Capybara-34B']]
18:26:55 - Layer 39/60 - RETAINED - -0.07127
----
Optimizing Layer 40/60 (slerp): 100%|█████████████| 4/4 [04:10<00:00, 62.57s/it]
[[1.0, 'jondurbin_bagel-dpo-34b-v0.2']]
18:32:47 - Layer 40/60 - RETAINED - -0.07127
----
Optimizing Layer 41/60 (slerp): 100%|█████████████| 4/4 [05:23<00:00, 80.96s/it]
[[1.0, 'jondurbin_bagel-dpo-34b-v0.2']]
18:39:44 - Layer 41/60 - RETAINED - -0.07127
----
Optimizing Layer 42/60 (slerp): 100%|█████████████| 4/4 [04:03<00:00, 60.87s/it]
[[1.0, 'jondurbin_bagel-dpo-34b-v0.2']]
18:45:31 - Layer 42/60 - RETAINED - -0.07127
----
Optimizing Layer 43/60 (slerp): 100%|█████████████| 4/4 [03:36<00:00, 54.22s/it]
[[1.0, 'jondurbin_bagel-dpo-34b-v0.2']]
18:50:34 - Layer 43/60 - RETAINED - -0.07127
----
Optimizing Layer 44/60 (slerp): 100%|█████████████| 4/4 [03:52<00:00, 58.18s/it]
[[1.0, 'jondurbin_bagel-dpo-34b-v0.2']]
18:55:44 - Layer 44/60 - RETAINED - -0.07127
----
Optimizing Layer 45/60 (slerp): 100%|█████████████| 4/4 [03:39<00:00, 54.92s/it]
[[1.0, 'jondurbin_bagel-dpo-34b-v0.2']]
19:00:39 - Layer 45/60 - RETAINED - -0.07127
----
Optimizing Layer 46/60 (slerp): 100%|█████████████| 4/4 [03:36<00:00, 54.06s/it]
[[1.0, 'jondurbin_bagel-dpo-34b-v0.2']]
19:05:24 - Layer 46/60 - RETAINED - -0.07127
----
Optimizing Layer 47/60 (slerp): 100%|█████████████| 4/4 [03:50<00:00, 57.54s/it]
[[1.0, 'jondurbin_bagel-dpo-34b-v0.2']]
19:10:28 - Layer 47/60 - RETAINED - -0.07127
----
Optimizing Layer 48/60 (slerp): 100%|█████████████| 4/4 [04:02<00:00, 60.62s/it]
[[1.0, 'jondurbin_bagel-dpo-34b-v0.2'], [0.2, 'NousResearch_Nous-Hermes-2-Yi-34B']]
19:15:45 - Layer 48/60 - RETAINED - -0.07127
----
Optimizing Layer 49/60 (slerp): 100%|█████████████| 4/4 [03:59<00:00, 59.77s/it]
[[1.0, 'jondurbin_bagel-dpo-34b-v0.2'], [0.4, 'NousResearch_Nous-Capybara-34B'], [0.8, 'NousResearch_Nous-Hermes-2-Yi-34B'], [0.6, 'SUSTech_SUS-Chat-34B']]
19:21:02 - Layer 49/60 - CHANGED - -0.07127 > -0.07407 - 3.9%
----
Optimizing Layer 50/60 (slerp): 100%|█████████████| 4/4 [03:53<00:00, 58.25s/it]
[[1.0, 'jondurbin_bagel-dpo-34b-v0.2'], [0.6, 'NousResearch_Nous-Capybara-34B'], [0.8, 'NousResearch_Nous-Hermes-2-Yi-34B'], [0.4, 'SUSTech_SUS-Chat-34B']]
19:26:11 - Layer 50/60 - CHANGED - -0.07407 > -0.07571 - 2.2%
----
Optimizing Layer 51/60 (slerp): 100%|█████████████| 4/4 [03:59<00:00, 59.91s/it]
[[1.0, 'jondurbin_bagel-dpo-34b-v0.2']]
19:31:30 - Layer 51/60 - RETAINED - -0.07571
----
Optimizing Layer 52/60 (slerp): 100%|█████████████| 4/4 [04:43<00:00, 70.77s/it]
[[1.0, 'jondurbin_bagel-dpo-34b-v0.2'], [0.8, 'NousResearch_Nous-Capybara-34B'], [0.8, 'NousResearch_Nous-Hermes-2-Yi-34B'], [0.6, 'SUSTech_SUS-Chat-34B']]
19:37:38 - Layer 52/60 - CHANGED - -0.07571 > -0.07660 - 1.2%
----
Optimizing Layer 53/60 (slerp): 100%|█████████████| 4/4 [04:26<00:00, 66.68s/it]
[[1.0, 'jondurbin_bagel-dpo-34b-v0.2'], [0.2, 'NousResearch_Nous-Capybara-34B'], [0.4, 'NousResearch_Nous-Hermes-2-Yi-34B'], [0.8, 'SUSTech_SUS-Chat-34B']]
19:43:27 - Layer 53/60 - CHANGED - -0.07660 > -0.07717 - 0.8%
----
Optimizing Layer 54/60 (slerp): 100%|█████████████| 4/4 [04:49<00:00, 72.34s/it]
[[1.0, 'jondurbin_bagel-dpo-34b-v0.2'], [0.4, 'NousResearch_Nous-Capybara-34B'], [0.8, 'NousResearch_Nous-Hermes-2-Yi-34B'], [0.8, 'SUSTech_SUS-Chat-34B']]
19:50:18 - Layer 54/60 - CHANGED - -0.07717 > -0.07775 - 0.7%
----
Optimizing Layer 55/60 (slerp): 100%|█████████████| 4/4 [04:12<00:00, 63.01s/it]
[[1.0, 'jondurbin_bagel-dpo-34b-v0.2'], [0.8, 'NousResearch_Nous-Hermes-2-Yi-34B'], [0.8, 'SUSTech_SUS-Chat-34B']]
19:56:01 - Layer 55/60 - CHANGED - -0.07775 > -0.07923 - 1.9%
----
Optimizing Layer 56/60 (slerp): 100%|█████████████| 4/4 [03:56<00:00, 59.03s/it]
[[1.0, 'jondurbin_bagel-dpo-34b-v0.2'], [0.8, 'NousResearch_Nous-Capybara-34B']]
20:01:25 - Layer 56/60 - RETAINED - -0.07923
----
Optimizing Layer 57/60 (slerp): 100%|█████████████| 4/4 [04:07<00:00, 61.99s/it]
[[1.0, 'jondurbin_bagel-dpo-34b-v0.2'], [0.8, 'NousResearch_Nous-Capybara-34B']]
20:06:54 - Layer 57/60 - RETAINED - -0.07923
----
Optimizing Layer 58/60 (slerp): 100%|█████████████| 4/4 [03:55<00:00, 58.84s/it]
[[1.0, 'jondurbin_bagel-dpo-34b-v0.2'], [0.6, 'NousResearch_Nous-Capybara-34B']]
20:12:09 - Layer 58/60 - RETAINED - -0.07923
----
Optimizing Layer 59/60 (slerp): 100%|█████████████| 4/4 [03:27<00:00, 51.80s/it]
[[1.0, 'jondurbin_bagel-dpo-34b-v0.2'], [0.8, 'NousResearch_Nous-Hermes-2-Yi-34B']]
20:16:49 - Layer 59/60 - RETAINED - -0.07923
----
Optimizing Layer 60/60 (slerp): 100%|█████████████| 4/4 [04:01<00:00, 60.29s/it]
[[1.0, 'jondurbin_bagel-dpo-34b-v0.2']]
20:22:08 - Layer 60/60 - RETAINED - -0.07923
----
Optimizing Header: 100%|██████████████████████████| 4/4 [03:49<00:00, 57.30s/it]
[[1.0, 'jondurbin_bagel-dpo-34b-v0.2'], [0.6, 'NousResearch_Nous-Capybara-34B'], [0.8, 'NousResearch_Nous-Hermes-2-Yi-34B'], [0.6, 'SUSTech_SUS-Chat-34B']]
20:26:56 - Header - CHANGED - -0.07923 > -0.07981 - 0.7%
-----------------------------------------------------------------------------------------------------
| Type | Phrase | Context | Raw Prob* | Used Prob** | Change |
-----------------------------------------------------------------------------------------------------
| BAD | anticipation | Her body quivers with | 0.00000% | 0.00% | +0.00% |
| BAD | anticipation | The atmosphere is thic.. | 0.00000% | 0.00% | +0.00% |
| BAD | unwavering | Filled with an | 0.00000% | 0.00% | +0.00% |
| BAD | determination | Her eyes were filled w.. | 0.00000% | 0.00% | -0.00% |
| BAD | determination | Her stubbornness only .. | 0.00000% | 0.00% | +0.00% |
| BAD | whisper | Her voice barely above.. | 0.00000% | 0.00% | +0.00% |
| BAD | spine | shivers down her | 0.00000% | 0.00% | +0.00% |
| BAD | sends shivers | The thrill of the act | 0.00000% | 0.00% | +0.00% |
| BAD | ministrations | She moans and twitches.. | 0.00004% | 0.00% | -0.00% |
| BAD | legs | wraps her | 0.00000% | 0.00% | -0.00% |
| BAD | imposing figure | He had an | 0.00000% | 0.00% | -0.00% |
| BAD | shared challenges | Their bond strengthene.. | 0.00001% | 0.00% | +0.00% |
| BAD | bond | forged a | 0.00005% | 0.00% | -0.00% |
| BAD | bond | an unspoken | 0.00010% | 0.00% | +0.00% |
| BAD | enhance our expe.. | I'm excited to see how | 0.00000% | 0.00% | +0.00% |
| BAD | sense of vulnera.. | create a | 0.00000% | 0.00% | -0.00% |
| BAD | dimensions of in.. | explore new | 0.00000% | 0.00% | +0.00% |
| BAD | deepening our co.. | while | 0.00000% | 0.00% | -0.00% |
| BAD | shared experiences | through | 0.00001% | 0.00% | +0.00% |
| BAD | societal expecta.. | that transcend | 0.00000% | 0.00% | -0.00% |
| BAD | conventional bou.. | that defy | 0.00000% | 0.00% | +0.00% |
| BAD | conventional bou.. | and defy | 0.00000% | 0.00% | +0.00% |
| BAD | open communication | an environment | 0.00000% | 0.00% | +0.00% |
| BAD | emotional vulner.. | an environment | 0.00000% | 0.00% | +0.00% |
| BAD | heightens our co.. | touch and the anticipa.. | 0.00000% | 0.00% | -0.00% |
| BAD | sensations you'r.. | I'm enjoying | 0.00000% | 0.00% | +0.00% |
| BAD | is truly arousing | attention to detail | 0.00000% | 0.00% | +0.00% |
| BAD | is truly arousing | way you explore my body | 0.00000% | 0.00% | +0.00% |
| BAD | challenge presen.. | my resolve unwavering .. | 0.00000% | 0.00% | +0.00% |
| BAD | humble vessel | surrendering to the ex.. | 0.00000% | 0.00% | +0.00% |
| BAD | bond | cherishing the unique | 0.00019% | 0.00% | +0.00% |
| BAD | bond | special | 0.00014% | 0.00% | -0.00% |
| BAD | grows stronger w.. | bond | 0.00000% | 0.00% | -0.00% |
| BAD | that cannot be b.. | bond | 0.00000% | 0.00% | -0.00% |
| BAD | becomes unbreaka.. | bond | 0.00000% | 0.00% | +0.00% |
| BAD | grew stronger wi.. | bond | 0.00000% | 0.00% | +0.00% |
| GOOD | The apple is in .. | Question: If I'm in th.. | 7.81435% | 7.81% | +7.81% |
------------------------------------------------------------------------------------------------------
| Totals | 7.81% | 7.82% | 7.81% |
------------------------------------------------------------------------------------------------------
* = Unweighted, raw probability - ** = Probability after weight adjustments
-------- MERGE COMPOSITION ---------
jondurbin_bagel-dpo-34b-v0.2: 0.49
NousResearch_Nous-Hermes-2-Yi-34B: 0.24
SUSTech_SUS-Chat-34B: 0.14
NousResearch_Nous-Capybara-34B: 0.13
20:28:04 - Saving model to ./mm-output...
20:28:48 - Copying tokenizer files to ./mm-output...
Skipped added_tokens.json (not found)
Copied tokenizer.model
Copied special_tokens_map.json
Copied tokenizer_config.json
Skipped vocab.json (not found)
Skipped merges.txt (not found)
20:28:48 - Model and tokenizer files saved successfully.