content large_stringlengths 3 20.5k | url large_stringlengths 53 192 β | branch large_stringclasses 4
values | source large_stringclasses 51
values | embeddings listlengths 384 384 | score float64 -0.21 0.65 |
|---|---|---|---|---|---|
number of elements in all processes participating in the collective. Parameters tensor (Tensor) β Data to be sent if src is the rank of current process, and tensor to be used to save received data otherwise. src (int) β Source rank on global process group (regardless of group argument). group (ProcessGroup, optional) β... | https://github.com/NousResearch/hermes-agent/blob/main/website/docs/user-guide/skills/optional/mlops/mlops-pytorch-fsdp.md | main | hermes-agent | [
-0.0232499148696661,
-0.08060023933649063,
-0.04737663269042969,
0.1147671490907669,
0.06563706696033478,
-0.018727710470557213,
0.043897733092308044,
-0.06506919860839844,
-0.0005893419147469103,
-0.04092475771903992,
-0.03665172681212425,
-0.012613904662430286,
-0.06486330926418304,
0.02... | 0.064045 |
be bitwise identical in all processes. Complex tensors are supported. Parameters tensor (Tensor) β Input and output of the collective. The function operates in-place. op (optional) β One of the values from torch.distributed.ReduceOp enum. Specifies an operation used for element-wise reductions. group (ProcessGroup, opt... | https://github.com/NousResearch/hermes-agent/blob/main/website/docs/user-guide/skills/optional/mlops/mlops-pytorch-fsdp.md | main | hermes-agent | [
-0.05235319957137108,
-0.06284406036138535,
-0.025397377088665962,
0.10389060527086258,
0.02849716693162918,
-0.07406490296125412,
-0.03744976222515106,
-0.03579748794436455,
-0.04505619779229164,
-0.07719788700342178,
-0.007775639183819294,
-0.017661256715655327,
-0.07929624617099762,
-0.... | 0.071691 |
device='cuda:0') # Rank 0 tensor([3, 4], device='cuda:1') # Rank 1 >>> dist.all\_gather(tensor\_list, tensor) >>> tensor\_list [tensor([1, 2], device='cuda:0'), tensor([3, 4], device='cuda:0')] # Rank 0 [tensor([1, 2], device='cuda:1'), tensor([3, 4], device='cuda:1')] # Rank 1 >>> # All tensors below are of torch.cflo... | https://github.com/NousResearch/hermes-agent/blob/main/website/docs/user-guide/skills/optional/mlops/mlops-pytorch-fsdp.md | main | hermes-agent | [
-0.03084925375878811,
-0.06889285147190094,
0.002976774936541915,
0.016788078472018242,
0.024431608617305756,
-0.0114488136023283,
0.031253620982170105,
-0.02062061056494713,
-0.07790343463420868,
-0.08854097127914429,
0.006077332887798548,
-0.08116807043552399,
-0.06686834245920181,
0.062... | 0.071154 |
the output of the collective will be populated into the input object\_list. If the calling rank is not part of the group, the passed in object\_list will be unmodified. Note Note that this API differs slightly from the all\_gather() collective since it does not provide an async\_op handle and thus will be a blocking ca... | https://github.com/NousResearch/hermes-agent/blob/main/website/docs/user-guide/skills/optional/mlops/mlops-pytorch-fsdp.md | main | hermes-agent | [
-0.09137289226055145,
-0.06780523806810379,
-0.09533132612705231,
0.10293565690517426,
0.02201257087290287,
-0.06506326049566269,
-0.026954591274261475,
-0.023083796724677086,
-0.04170699045062065,
-0.07989270240068436,
-0.018443455919623375,
-0.036820851266384125,
0.01633528247475624,
-0.... | 0.075904 |
non-dst ranks. (default is None) dst (int, optional) β Destination rank on global process group (regardless of group argument). (If both dst and group\_dst are None, default is global rank 0) group (Optional[ProcessGroup]) β (ProcessGroup, optional): The process group to work on. If None, the default process group will... | https://github.com/NousResearch/hermes-agent/blob/main/website/docs/user-guide/skills/optional/mlops/mlops-pytorch-fsdp.md | main | hermes-agent | [
-0.12377478182315826,
-0.07219783216714859,
-0.0701964944601059,
0.09313932061195374,
0.015565569512546062,
-0.08066761493682861,
-0.00464208098128438,
0.022026866674423218,
-0.012666745111346245,
-0.07106038182973862,
-0.030049849301576614,
-0.0034680459648370743,
-0.010269026271998882,
-... | 0.063028 |
scatter\_list, src=0) >>> # Rank i gets scatter\_list[i]. >>> output\_tensor tensor([1., 1.], device='cuda:0') # Rank 0 tensor([5., 5.], device='cuda:1') # Rank 1 torch.distributed.scatter\_object\_list(scatter\_object\_output\_list, scatter\_object\_input\_list=None, src=None, group=None, group\_src=None)[source]# Sca... | https://github.com/NousResearch/hermes-agent/blob/main/website/docs/user-guide/skills/optional/mlops/mlops-pytorch-fsdp.md | main | hermes-agent | [
0.014957576058804989,
-0.09571264684200287,
0.003228964051231742,
0.05688124895095825,
0.0956227257847786,
-0.0038352764677256346,
0.0463809110224247,
-0.023369133472442627,
-0.07491195201873779,
-0.06531417369842529,
-0.007071241736412048,
-0.04238872230052948,
-0.01460344810038805,
0.054... | 0.04669 |
be reduced and scattered. Its size should be output tensor size times the world size. The input tensor can have one of the following shapes: (i) a concatenation of the output tensors along the primary dimension, or (ii) a stack of the output tensors along the primary dimension. For definition of βconcatenationβ, see to... | https://github.com/NousResearch/hermes-agent/blob/main/website/docs/user-guide/skills/optional/mlops/mlops-pytorch-fsdp.md | main | hermes-agent | [
-0.02627447061240673,
-0.06649427860975266,
-0.043835241347551346,
0.08653969317674637,
-0.016346141695976257,
-0.04611269757151604,
-0.038760483264923096,
-0.01510393712669611,
-0.06904304027557373,
-0.14014901220798492,
0.006434649229049683,
-0.0016472924035042524,
-0.11517505347728729,
... | 0.059101 |
21, 22, 23, 24]) # Rank 2 tensor([30, 31, 32, 33, 34, 35, 36]) # Rank 3 >>> input\_splits [2, 2, 1, 1] # Rank 0 [3, 2, 2, 2] # Rank 1 [2, 1, 1, 1] # Rank 2 [2, 2, 2, 1] # Rank 3 >>> output\_splits [2, 3, 2, 2] # Rank 0 [2, 2, 1, 2] # Rank 1 [1, 2, 1, 2] # Rank 2 [1, 2, 1, 1] # Rank 3 >>> output = ... >>> dist.all\_to\_... | https://github.com/NousResearch/hermes-agent/blob/main/website/docs/user-guide/skills/optional/mlops/mlops-pytorch-fsdp.md | main | hermes-agent | [
0.009482978843152523,
-0.0498102605342865,
0.03525576740503311,
-0.04572959616780281,
0.0794476643204689,
0.06553002446889877,
-0.0025484052021056414,
-0.036386776715517044,
-0.12345011532306671,
-0.02923835813999176,
-0.04015367478132248,
0.026731878519058228,
-0.009599483571946621,
0.072... | 0.00423 |
1, 2] # Rank 2 [1, 2, 1, 1] # Rank 3 >>> input = list(input.split(input\_splits)) >>> input [tensor([0, 1]), tensor([2, 3]), tensor([4]), tensor([5])] # Rank 0 [tensor([10, 11, 12]), tensor([13, 14]), tensor([15, 16]), tensor([17, 18])] # Rank 1 [tensor([20, 21]), tensor([22]), tensor([23]), tensor([24])] # Rank 2 [ten... | https://github.com/NousResearch/hermes-agent/blob/main/website/docs/user-guide/skills/optional/mlops/mlops-pytorch-fsdp.md | main | hermes-agent | [
-0.005507820285856724,
-0.050229936838150024,
0.03099771961569786,
-0.03655598312616348,
0.05636623874306679,
0.06222480908036232,
0.029185771942138672,
-0.05897555872797966,
-0.09014492481946945,
-0.040583714842796326,
-0.04388844221830368,
0.01269625686109066,
-0.012836411595344543,
0.07... | 0.044139 |
purposes, this barrier can be inserted before the applicationβs collective calls to check if any ranks are desynchronized. Note Note that this collective is only supported with the GLOO backend. Parameters group (ProcessGroup, optional) β The process group to work on. If None, the default process group will be used. ti... | https://github.com/NousResearch/hermes-agent/blob/main/website/docs/user-guide/skills/optional/mlops/mlops-pytorch-fsdp.md | main | hermes-agent | [
-0.08768707513809204,
-0.032127745449543,
-0.03262361139059067,
0.07282301038503647,
0.06622882187366486,
-0.023800073191523552,
-0.010653682984411716,
0.00023408295237459242,
-0.04247937723994255,
-0.011631148867309093,
-0.013035820797085762,
-0.011665187776088715,
-0.02558738924562931,
-... | 0.101498 |
CPU-GPU work (e.g. sending GPU tensors with GLOO), fut.done() returns true when tensors have arrived on respective nodes, but not yet necessarily synched on respective GPUs (similarly to GPU work). get\_future\_result(self: torch.\_C.\_distributed\_c10d.Work) β torch.Future# Returns A torch.futures.Future object of int... | https://github.com/NousResearch/hermes-agent/blob/main/website/docs/user-guide/skills/optional/mlops/mlops-pytorch-fsdp.md | main | hermes-agent | [
-0.0645373985171318,
-0.02410239912569523,
-0.05730465427041054,
0.05150628462433815,
0.007074833381921053,
-0.08842378854751587,
-0.009722679853439331,
-0.014588909223675728,
0.00237565697170794,
-0.047130219638347626,
-0.02277076244354248,
-0.009870161302387714,
-0.09363708645105362,
-0.... | 0.113989 |
be used >>> store = dist.TCPStore("127.0.0.1", 0, 1, True, timedelta(seconds=30)) >>> store.add("first\_key", 1) >>> store.add("first\_key", 6) >>> # Should return 7 >>> store.get("first\_key") append(self: torch.\_C.\_distributed\_c10d.Store, arg0: str, arg1: str) β None# Append the key-value pair into the store based... | https://github.com/NousResearch/hermes-agent/blob/main/website/docs/user-guide/skills/optional/mlops/mlops-pytorch-fsdp.md | main | hermes-agent | [
-0.048452433198690414,
0.0332789309322834,
-0.06883280724287033,
0.039025768637657166,
-0.13454069197177887,
-0.011239366605877876,
0.03902590274810791,
0.05119689926505089,
0.02027410827577114,
0.0004910866846330464,
0.08475978672504425,
-0.057937029749155045,
-0.0012293142499402165,
-0.0... | -0.002978 |
present in the store, the function will wait for timeout, which is defined when initializing the store, before throwing an exception. Parameters key (str) β The function will return the value associated with this key. Returns Value associated with key if key is in the store. Example::>>> import torch.distributed as dis... | https://github.com/NousResearch/hermes-agent/blob/main/website/docs/user-guide/skills/optional/mlops/mlops-pytorch-fsdp.md | main | hermes-agent | [
-0.03501158207654953,
0.04979012906551361,
-0.05429313704371452,
0.06341371685266495,
-0.07840141654014587,
-0.02334684692323208,
0.01083379052579403,
0.03281977400183678,
0.03734315559267998,
-0.02005179598927498,
0.08511684089899063,
-0.03603661060333252,
-0.06867107003927231,
-0.0545287... | 0.033966 |
pair into the store based on the supplied key and value. If key already exists in the store, it will overwrite the old value with the new supplied value. Parameters key (str) β The key to be added to the store. value (str) β The value associated with key to be added to the store. Example::>>> import torch.distributed a... | https://github.com/NousResearch/hermes-agent/blob/main/website/docs/user-guide/skills/optional/mlops/mlops-pytorch-fsdp.md | main | hermes-agent | [
-0.05110529437661171,
0.018181907013058662,
-0.06515256315469742,
0.08236479759216309,
-0.09840673953294754,
0.023447534069418907,
0.02132314443588257,
0.030631257221102715,
0.03202510625123978,
-0.008926081471145153,
0.08017702400684357,
-0.005051999818533659,
-0.05249667912721634,
-0.054... | 0.005064 |
timedelta(seconds=300) wait\_for\_workers (bool, optional) β Whether to wait for all the workers to connect with the server store. This is only applicable when world\_size is a fixed value. Default is True. multi\_tenant (bool, optional) β If True, all TCPStore instances in the current process with the same host/port w... | https://github.com/NousResearch/hermes-agent/blob/main/website/docs/user-guide/skills/optional/mlops/mlops-pytorch-fsdp.md | main | hermes-agent | [
-0.05070404335856438,
0.01789660006761551,
-0.09282218664884567,
0.009239443577826023,
-0.08814667165279388,
-0.0316971093416214,
-0.031956467777490616,
0.003458064515143633,
0.010263554751873016,
0.0038502460811287165,
0.024742841720581055,
-0.021775245666503906,
-0.06744859367609024,
-0.... | 0.019301 |
or torch.autograd.profiler to profile collective communication and point-to-point communication APIs mentioned here. All out-of-the-box backends (gloo, nccl, mpi) are supported and collective communication usage will be rendered as expected in profiling output/traces. Profiling your code is the same as any regular torc... | https://github.com/NousResearch/hermes-agent/blob/main/website/docs/user-guide/skills/optional/mlops/mlops-pytorch-fsdp.md | main | hermes-agent | [
-0.021947775036096573,
-0.08269746601581573,
-0.0883321613073349,
0.05653313919901848,
0.04625355452299118,
-0.07595434039831161,
-0.01755419187247753,
0.04632340744137764,
-0.0682235062122345,
-0.08372107148170471,
-0.04901134595274925,
-0.028190433979034424,
-0.04898006469011307,
0.05659... | 0.057398 |
how to develop a third-party backend through C++ Extension, please refer to Tutorials - Custom C++ and CUDA Extensions and test/cpp\_extensions/cpp\_c10d\_extension.cpp. The capability of third-party backends are decided by their own implementations. The new backend derives from c10d::ProcessGroup and registers the bac... | https://github.com/NousResearch/hermes-agent/blob/main/website/docs/user-guide/skills/optional/mlops/mlops-pytorch-fsdp.md | main | hermes-agent | [
-0.10436389595270157,
-0.13629426062107086,
-0.08647177368402481,
0.018181584775447845,
0.014885542914271355,
0.0035829851403832436,
-0.03780164569616318,
0.06144474819302559,
-0.0607505664229393,
-0.062469810247421265,
0.04513596370816231,
-0.01859935186803341,
-0.0601041354238987,
-0.017... | 0.095684 |
script. From PyTorch 2.0.0 onwards, the dashed --local-rank is preferred over the previously used underscored --local\_rank. For backward compatibility, it may be necessary for users to handle both cases in their argument parsing code. This means including both "--local-rank" and "--local\_rank" in the argument parser.... | https://github.com/NousResearch/hermes-agent/blob/main/website/docs/user-guide/skills/optional/mlops/mlops-pytorch-fsdp.md | main | hermes-agent | [
-0.06572063267230988,
-0.07197018712759018,
-0.02862890437245369,
0.02536594681441784,
0.00421390775591135,
0.06310389935970306,
-0.027135498821735382,
0.06395374238491058,
-0.0927455723285675,
-0.07860375195741653,
0.0062096016481518745,
0.0159305352717638,
-0.04790542274713516,
0.0453166... | -0.105409 |
in each case. Monitored Barrier# As of v1.10, torch.distributed.monitored\_barrier() exists as an alternative to torch.distributed.barrier() which fails with helpful information about which rank may be faulty when crashing, i.e. not all ranks calling into torch.distributed.monitored\_barrier() within the provided timeo... | https://github.com/NousResearch/hermes-agent/blob/main/website/docs/user-guide/skills/optional/mlops/mlops-pytorch-fsdp.md | main | hermes-agent | [
-0.04584153741598129,
-0.05222296342253685,
-0.004008662421256304,
0.06916750222444534,
0.05013131722807884,
-0.06890535354614258,
-0.01559380255639553,
0.045616935938596725,
-0.03602976351976395,
0.035348113626241684,
-0.047572724521160126,
-0.00009409273479832336,
-0.044469885528087616,
... | 0.056296 |
overlap time: 2234674 In addition, TORCH\_DISTRIBUTED\_DEBUG=INFO enhances crash logging in torch.nn.parallel.DistributedDataParallel() due to unused parameters in the model. Currently, find\_unused\_parameters=True must be passed into torch.nn.parallel.DistributedDataParallel() initialization if there are parameters t... | https://github.com/NousResearch/hermes-agent/blob/main/website/docs/user-guide/skills/optional/mlops/mlops-pytorch-fsdp.md | main | hermes-agent | [
-0.05034720525145531,
-0.09047147631645203,
0.015558402985334396,
0.09083651006221771,
0.023920010775327682,
-0.04307921603322029,
-0.05254882946610451,
0.04779958724975586,
-0.049837708473205566,
-0.019259212538599968,
0.010822992771863937,
0.01695289835333824,
-0.06413699686527252,
-0.04... | -0.013662 |
across ranks. Got shapes: 10 [ torch.LongTensor{1} ] Note For fine-grained control of the debug level during runtime the functions torch.distributed.set\_debug\_level(), torch.distributed.set\_debug\_level\_from\_env(), and torch.distributed.get\_debug\_level() can also be used. In addition, TORCH\_DISTRIBUTED\_DEBUG=D... | https://github.com/NousResearch/hermes-agent/blob/main/website/docs/user-guide/skills/optional/mlops/mlops-pytorch-fsdp.md | main | hermes-agent | [
-0.0627872571349144,
-0.07383895665407181,
-0.02821418084204197,
0.07494593411684036,
0.06878853589296341,
-0.11343178898096085,
-0.03538504242897034,
0.03360539302229881,
-0.0394514799118042,
-0.030510082840919495,
-0.004934542812407017,
-0.042250845581293106,
-0.11203189939260483,
-0.012... | 0.066118 |
neither is specified, init\_method is assumed to be βenv://β. Parameters backend (str or Backend, optional) β The backend to use. Depending on build-time configurations, valid values include mpi, gloo, nccl, ucc, xccl or one that is registered by a third-party plugin. Since 2.6, if backend is not provided, c10d will us... | https://github.com/NousResearch/hermes-agent/blob/main/website/docs/user-guide/skills/optional/mlops/mlops-pytorch-fsdp.md | main | hermes-agent | [
-0.08745770156383514,
-0.028595497831702232,
-0.12740302085876465,
-0.03125748410820961,
0.013364668935537338,
-0.054345306009054184,
-0.0016091926954686642,
0.08053028583526611,
-0.02663581445813179,
-0.07252253592014313,
0.023629486560821533,
-0.048018842935562134,
0.03702149912714958,
-... | 0.16507 |
will be created. The gloo backend will be used for collectives with CPU tensors and the nccl backend will be used for collectives with CUDA tensors. A custom backend can be specified by passing in a string with format β<device\_type>:<backend\_name>,<device\_type>:<backend\_name>β, e.g. βcpu:gloo,cuda:custom\_backendβ.... | https://github.com/NousResearch/hermes-agent/blob/main/website/docs/user-guide/skills/optional/mlops/mlops-pytorch-fsdp.md | main | hermes-agent | [
-0.0844976007938385,
-0.06406056880950928,
-0.09941200166940689,
0.01963551715016365,
-0.032514311373233795,
-0.06530702859163284,
0.007554008159786463,
0.0421644002199173,
-0.07983724027872086,
-0.043980684131383896,
0.036351632326841354,
-0.05696755647659302,
-0.0376870222389698,
-0.0064... | 0.210632 |
Note that multicast address is not supported anymore in the latest distributed package. group\_name is deprecated as well. import torch.distributed as dist # Use address of one of the machines dist.init\_process\_group(backend, init\_method='tcp://10.1.1.20:23456', rank=args.rank, world\_size=4) Shared file-system init... | https://github.com/NousResearch/hermes-agent/blob/main/website/docs/user-guide/skills/optional/mlops/mlops-pytorch-fsdp.md | main | hermes-agent | [
-0.07138971984386444,
-0.048002034425735474,
-0.031792402267456055,
-0.027148831635713577,
0.028514163568615913,
-0.0475170724093914,
-0.00317126652225852,
0.020007755607366562,
-0.014555679634213448,
-0.004100988153368235,
0.059763628989458084,
0.038442619144916534,
-0.03327396139502525,
... | 0.028086 |
world) and require all processes to enter the distributed function call. However, some workloads can benefit from more fine-grained communication. This is where distributed groups come into play. new\_group() function can be used to create new groups, with arbitrary subsets of all processes. It returns an opaque group ... | https://github.com/NousResearch/hermes-agent/blob/main/website/docs/user-guide/skills/optional/mlops/mlops-pytorch-fsdp.md | main | hermes-agent | [
-0.09507041424512863,
-0.11101400852203369,
-0.08050694316625595,
0.04879363626241684,
0.01708521693944931,
-0.09791669249534607,
-0.025393327698111534,
-0.05160247161984444,
0.048476435244083405,
0.01736067421734333,
0.01593725010752678,
0.04841620847582817,
-0.052170202136039734,
-0.0193... | 0.125921 |
all ranks follow the same global creation order. torch.distributed.get\_group\_rank(group, global\_rank)[source]# Translate a global rank into a group rank. global\_rank must be part of group otherwise this raises RuntimeError. Parameters group (ProcessGroup) β ProcessGroup to find the relative rank. global\_rank (int)... | https://github.com/NousResearch/hermes-agent/blob/main/website/docs/user-guide/skills/optional/mlops/mlops-pytorch-fsdp.md | main | hermes-agent | [
-0.06693760305643082,
-0.0867282971739769,
-0.02877121977508068,
0.0706389769911766,
0.005851468537002802,
-0.030548779293894768,
0.02171037532389164,
0.026557641103863716,
-0.01962416060268879,
-0.013308502733707428,
0.03595299273729324,
-0.021697163581848145,
0.01292983815073967,
-0.0027... | -0.011618 |
unused parameters, as DDP will not search graph in each iteration to detect unused parameters when static\_graph is set to be True. To check whether you can set static\_graph to be True, one way is to check ddp logging data at the end of your previous model training, if ddp\_logging\_data.get("can\_set\_static\_graph")... | https://github.com/NousResearch/hermes-agent/blob/main/website/docs/user-guide/skills/optional/mlops/mlops-pytorch-fsdp.md | main | hermes-agent | [
-0.07057483494281769,
-0.058853067457675934,
-0.04383613169193268,
0.05747269466519356,
0.036991868168115616,
-0.05203395336866379,
-0.01502966694533825,
-0.001818413962610066,
-0.11987762153148651,
-0.05422453209757805,
0.056407488882541656,
0.01583169773221016,
-0.026248980313539505,
-0.... | 0.007211 |
{/\* This page is auto-generated from the skill's SKILL.md by website/scripts/generate-skill-docs.py. Edit the source SKILL.md, not this page. \*/} # Pinecone Managed vector database for production AI applications. Fully managed, auto-scaling, with hybrid search (dense + sparse), metadata filtering, and namespaces. Low... | https://github.com/NousResearch/hermes-agent/blob/main/website/docs/user-guide/skills/optional/mlops/mlops-pinecone.md | main | hermes-agent | [
-0.02619091048836708,
-0.09438590705394745,
-0.042448390275239944,
0.07319656014442444,
-0.004892834462225437,
-0.05758222937583923,
0.011083443649113178,
0.0032302497420459986,
-0.1148560494184494,
0.037428904324769974,
-0.006759015377610922,
-0.004342485684901476,
0.04241897910833359,
0.... | 0.165538 |
# Partition data by namespace index.upsert( vectors=[{"id": "vec1", "values": [...]}], namespace="user-123" ) # Query specific namespace results = index.query( vector=[...], namespace="user-123", top\_k=5 ) # List namespaces stats = index.describe\_index\_stats() print(stats['namespaces']) ``` ## Hybrid search (dense +... | https://github.com/NousResearch/hermes-agent/blob/main/website/docs/user-guide/skills/optional/mlops/mlops-pinecone.md | main | hermes-agent | [
-0.02786148712038994,
0.0036190964747220278,
-0.023199720308184624,
0.005427287891507149,
0.0025040842592716217,
-0.027749691158533096,
0.04609455168247223,
0.024097982794046402,
-0.028677280992269516,
-0.011292085982859135,
0.007341209799051285,
-0.01033847127109766,
0.08038230240345001,
... | 0.020512 |
{/\* This page is auto-generated from the skill's SKILL.md by website/scripts/generate-skill-docs.py. Edit the source SKILL.md, not this page. \*/} # Pytorch Lightning High-level PyTorch framework with Trainer class, automatic distributed training (DDP/FSDP/DeepSpeed), callbacks system, and minimal boilerplate. Scales ... | https://github.com/NousResearch/hermes-agent/blob/main/website/docs/user-guide/skills/optional/mlops/mlops-pytorch-lightning.md | main | hermes-agent | [
-0.10767008364200592,
-0.05455709248781204,
-0.05271847918629646,
0.03337519243359566,
-0.03637702763080597,
-0.03836256265640259,
-0.007787882816046476,
-0.018578434363007545,
-0.10838907957077026,
-0.08348964899778366,
0.03098769299685955,
0.02224419079720974,
-0.03250996395945549,
-0.03... | 0.048306 |
L.Trainer( accelerator='gpu', devices=8, strategy='ddp' # Or 'fsdp', 'deepspeed' ) trainer.fit(model, train\_loader) ``` \*\*Launch\*\*: ```bash # Single command, Lightning handles the rest python train.py ``` \*\*No changes needed\*\*: - Automatic data distribution - Gradient synchronization - Multi-node support (just... | https://github.com/NousResearch/hermes-agent/blob/main/website/docs/user-guide/skills/optional/mlops/mlops-pytorch-lightning.md | main | hermes-agent | [
-0.06573331356048584,
-0.07729405164718628,
-0.03353462740778923,
0.08754515647888184,
0.031263574957847595,
0.007848132401704788,
-0.048012834042310715,
-0.010032816790044308,
-0.02825060859322548,
-0.1324477195739746,
-0.010545840486884117,
-0.0267501063644886,
-0.032385241240262985,
-0.... | -0.008368 |
{/\* This page is auto-generated from the skill's SKILL.md by website/scripts/generate-skill-docs.py. Edit the source SKILL.md, not this page. \*/} # Optimizing Attention Flash Optimizes transformer attention with Flash Attention for 2-4x speedup and 10-20x memory reduction. Use when training/running transformers with ... | https://github.com/NousResearch/hermes-agent/blob/main/website/docs/user-guide/skills/optional/mlops/mlops-flash-attention.md | main | hermes-agent | [
-0.09611286967992783,
-0.023308927193284035,
-0.06785055249929428,
0.024696048349142075,
0.006474462803453207,
-0.06205929070711136,
-0.014715440571308136,
0.04901721701025963,
-0.07661545276641846,
-0.04726275056600571,
0.03362560272216797,
0.04093967750668526,
-0.03864523768424988,
0.013... | 0.100863 |
(out\_flash - out\_standard).abs().max() print(f"Max difference: {diff:.6f}") # Should be <1e-3 for float16 ``` ### Workflow 2: Use flash-attn library for advanced features For multi-query attention, sliding window, or H100 FP8. Copy this checklist: ``` flash-attn Library Setup: - [ ] Step 1: Install flash-attn library... | https://github.com/NousResearch/hermes-agent/blob/main/website/docs/user-guide/skills/optional/mlops/mlops-flash-attention.md | main | hermes-agent | [
-0.02720055729150772,
-0.029690928757190704,
-0.0740356594324112,
0.011223859153687954,
0.04947574436664581,
-0.08243251591920853,
-0.01053479965776205,
0.087234266102314,
-0.05846342071890831,
-0.09894989430904388,
-0.005605859216302633,
-0.025013528764247894,
-0.07513061165809631,
0.0090... | -0.021229 |
\*\*Memory-efficient attention\*\*: CPU inference (Flash Attention needs GPU) ## Common issues \*\*Issue: ImportError: cannot import flash\_attn\*\* Install with no-build-isolation flag: ```bash pip install flash-attn --no-build-isolation ``` Or install CUDA toolkit first: ```bash conda install cuda -c nvidia pip insta... | https://github.com/NousResearch/hermes-agent/blob/main/website/docs/user-guide/skills/optional/mlops/mlops-flash-attention.md | main | hermes-agent | [
-0.04869129881262779,
-0.08681703358888626,
-0.08767636120319366,
0.04812156781554222,
0.04445984959602356,
-0.059503573924303055,
-0.003364305943250656,
0.028481746092438698,
-0.06651538610458374,
-0.08144918829202652,
0.066827692091465,
-0.041861630976200104,
-0.09914885461330414,
0.0460... | 0.040865 |
{/\* This page is auto-generated from the skill's SKILL.md by website/scripts/generate-skill-docs.py. Edit the source SKILL.md, not this page. \*/} # Modal Serverless Gpu Serverless GPU cloud platform for running ML workloads. Use when you need on-demand GPU access without infrastructure management, deploying ML models... | https://github.com/NousResearch/hermes-agent/blob/main/website/docs/user-guide/skills/optional/mlops/mlops-modal.md | main | hermes-agent | [
-0.0775270015001297,
-0.033349111676216125,
-0.03982667624950409,
0.028253929689526558,
-0.020513866096735,
-0.07910609245300293,
0.002535657724365592,
-0.054832108318805695,
-0.09359151870012283,
-0.020641200244426727,
0.01927337981760502,
0.026240643113851547,
0.06082718074321747,
-0.025... | 0.094747 |
| 48GB | Recommended for inference (best cost/perf) | | `A100-40GB` | 40GB | Large model training | | `A100-80GB` | 80GB | Very large models | | `H100` | 80GB | Fastest, FP8 + Transformer Engine | | `H200` | 141GB | Auto-upgrade from H100, 4.8TB/s bandwidth | | `B200` | Latest | Blackwell architecture | ### GPU specifi... | https://github.com/NousResearch/hermes-agent/blob/main/website/docs/user-guide/skills/optional/mlops/mlops-modal.md | main | hermes-agent | [
-0.026049142703413963,
0.014853578060865402,
-0.055603813380002975,
0.025705765932798386,
0.008289948105812073,
-0.0655895546078682,
-0.07751845568418503,
0.12759777903556824,
-0.10532752424478531,
-0.06747415661811829,
-0.004851159639656544,
0.009126040153205395,
0.01095576211810112,
-0.0... | -0.015806 |
{/\* This page is auto-generated from the skill's SKILL.md by website/scripts/generate-skill-docs.py. Edit the source SKILL.md, not this page. \*/} # Simpo Training Simple Preference Optimization for LLM alignment. Reference-free alternative to DPO with better performance (+6.4 points on AlpacaEval 2.0). No reference m... | https://github.com/NousResearch/hermes-agent/blob/main/website/docs/user-guide/skills/optional/mlops/mlops-simpo.md | main | hermes-agent | [
-0.09760589152574539,
-0.029696503654122353,
-0.029436571523547173,
-0.027567461133003235,
-0.08708091825246811,
-0.0690104216337204,
-0.024655289947986603,
0.04973417520523071,
-0.09040474891662598,
0.055943090468645096,
0.022660160437226295,
-0.0005327860708348453,
0.013383409939706326,
... | 0.119233 |
Reduce from 2.0 ``` \*\*Issue: Model forgets capabilities\*\* Add SFT regularization: ```yaml sft\_weight: 0.1 # Add SFT loss component ``` \*\*Issue: Poor preference separation\*\* Increase beta and margin: ```yaml beta: 5.0 # Increase from 2.0 gamma\_beta\_ratio: 0.8 # Increase from 0.5 ``` \*\*Issue: OOM during trai... | https://github.com/NousResearch/hermes-agent/blob/main/website/docs/user-guide/skills/optional/mlops/mlops-simpo.md | main | hermes-agent | [
-0.028740910813212395,
0.006688350345939398,
0.017403407022356987,
0.040059320628643036,
0.026015320792794228,
-0.0645538792014122,
-0.025432130321860313,
0.07351778447628021,
-0.024067532271146774,
0.008106522262096405,
0.03465494140982628,
0.007433265447616577,
-0.04322925955057144,
-0.0... | 0.136 |
{/\* This page is auto-generated from the skill's SKILL.md by website/scripts/generate-skill-docs.py. Edit the source SKILL.md, not this page. \*/} # Tensorrt Llm Optimizes LLM inference with NVIDIA TensorRT for maximum throughput and lowest latency. Use for production deployment on NVIDIA GPUs (A100/H100), when you ne... | https://github.com/NousResearch/hermes-agent/blob/main/website/docs/user-guide/skills/optional/mlops/mlops-tensorrt-llm.md | main | hermes-agent | [
0.005771054420620203,
-0.06840888410806656,
-0.060806404799222946,
-0.02173180691897869,
-0.014909656718373299,
-0.02751338854432106,
-0.09170844405889511,
-0.00857512280344963,
-0.09118896722793579,
-0.03755958378314972,
-0.056139808148145676,
0.00935718696564436,
-0.05307643860578537,
0.... | 0.04852 |
model="meta-llama/Meta-Llama-3-405B", tensor\_parallel\_size=8, dtype="fp8" ) ``` ### Batch inference ```python # Process 100 prompts efficiently prompts = [f"Question {i}: ..." for i in range(100)] outputs = llm.generate( prompts, sampling\_params=SamplingParams(max\_tokens=200) ) # Automatic in-flight batching for ma... | https://github.com/NousResearch/hermes-agent/blob/main/website/docs/user-guide/skills/optional/mlops/mlops-tensorrt-llm.md | main | hermes-agent | [
0.03613070026040077,
-0.019358128309249878,
-0.07313771545886993,
-0.02019806206226349,
-0.020458940416574478,
-0.07053147256374359,
-0.03614870086312294,
0.034725453704595566,
-0.0678539127111435,
0.01295744813978672,
-0.07113479822874069,
-0.0653160959482193,
-0.04054131358861923,
-0.003... | 0.055938 |
{/\* This page is auto-generated from the skill's SKILL.md by website/scripts/generate-skill-docs.py. Edit the source SKILL.md, not this page. \*/} # Huggingface Accelerate Simplest distributed training API. 4 lines to add distributed support to any PyTorch script. Unified API for DeepSpeed/FSDP/Megatron/DDP. Automatic... | https://github.com/NousResearch/hermes-agent/blob/main/website/docs/user-guide/skills/optional/mlops/mlops-accelerate.md | main | hermes-agent | [
-0.12895074486732483,
-0.031149111688137054,
-0.03496117889881134,
0.016622889786958694,
-0.04567777365446091,
-0.07537508010864258,
-0.1033308282494545,
-0.008921063505113125,
-0.15247303247451782,
-0.021034175530076027,
0.044426847249269485,
-0.03453228622674942,
0.002998386276885867,
-0... | 0.114663 |
"reduce\_bucket\_size": 5e8 } } ``` \*\*Launch\*\*: ```bash accelerate launch --config\_file deepspeed\_config.json train.py ``` ### Workflow 4: FSDP (Fully Sharded Data Parallel) \*\*Enable FSDP\*\*: ```python from accelerate import Accelerator, FullyShardedDataParallelPlugin fsdp\_plugin = FullyShardedDataParallelPlu... | https://github.com/NousResearch/hermes-agent/blob/main/website/docs/user-guide/skills/optional/mlops/mlops-accelerate.md | main | hermes-agent | [
-0.018418412655591965,
0.06835427135229111,
-0.026821887120604515,
0.005171656608581543,
-0.020622139796614647,
-0.0773441344499588,
-0.04316771402955055,
0.055473629385232925,
-0.06650260835886002,
-0.03465558961033821,
0.04449363797903061,
-0.06521590054035187,
-0.015190155245363712,
-0.... | 0.05496 |
{/\* This page is auto-generated from the skill's SKILL.md by website/scripts/generate-skill-docs.py. Edit the source SKILL.md, not this page. \*/} # Guidance Control LLM output with regex and grammars, guarantee valid JSON/XML/code generation, enforce structured formats, and build multi-step workflows with Guidance - ... | https://github.com/NousResearch/hermes-agent/blob/main/website/docs/user-guide/skills/optional/mlops/mlops-guidance.md | main | hermes-agent | [
-0.04762668162584305,
0.030873602256178856,
-0.04723650962114334,
-0.0014072300400584936,
0.01375316921621561,
-0.04651413857936859,
-0.03227370232343674,
-0.03381424397230148,
-0.10897108912467957,
0.0367291122674942,
-0.015793099999427795,
-0.006000537425279617,
0.07051599770784378,
-0.0... | 0.086296 |
name="sentiment") # Multiple-choice selection lm += "Best answer: " + select( ["A) Paris", "B) London", "C) Berlin", "D) Madrid"], name="answer" ) print(lm["sentiment"]) # One of: positive, negative, neutral print(lm["answer"]) # One of: A, B, C, or D ``` ### 3. Token Healing Guidance automatically "heals" token bounda... | https://github.com/NousResearch/hermes-agent/blob/main/website/docs/user-guide/skills/optional/mlops/mlops-guidance.md | main | hermes-agent | [
-0.007653404027223587,
0.012807013466954231,
0.05564291030168533,
0.025379827246069908,
0.042751945555210114,
-0.02342250756919384,
0.06377928704023361,
0.01172563061118126,
-0.02589447610080242,
0.025393709540367126,
0.018433023244142532,
-0.07226773351430893,
0.04033666476607323,
-0.0421... | 0.069973 |
+= "Sentiment: " + select(["positive", "negative", "neutral"], name="sentiment") lm += "\nConfidence: " + gen("confidence", regex=r"[0-9]+", max\_tokens=3) + "%" print(f"Sentiment: {lm['sentiment']}") print(f"Confidence: {lm['confidence']}%") ``` ### Pattern 3: Multi-Step Reasoning ```python from guidance import models... | https://github.com/NousResearch/hermes-agent/blob/main/website/docs/user-guide/skills/optional/mlops/mlops-guidance.md | main | hermes-agent | [
-0.062043994665145874,
0.04397376626729965,
0.012482798658311367,
0.013424077071249485,
0.030175218358635902,
-0.0026322994381189346,
0.03968804329633713,
0.09174320101737976,
-0.016954738646745682,
-0.01728711649775505,
0.030423501506447792,
-0.0748315304517746,
0.08488892763853073,
-0.02... | 0.059257 |
slow lm += gen("name", regex=r"^(John|Jane)$", max\_tokens=10) ``` ## Comparison to Alternatives | Feature | Guidance | Instructor | Outlines | LMQL | |---------|----------|------------|----------|------| | Regex Constraints | β
Yes | β No | β
Yes | β
Yes | | Grammar Support | β
CFG | β No | β
CFG | β
CFG | | Pydantic ... | https://github.com/NousResearch/hermes-agent/blob/main/website/docs/user-guide/skills/optional/mlops/mlops-guidance.md | main | hermes-agent | [
0.004428866785019636,
-0.054756127297878265,
0.022922199219465256,
0.034677792340517044,
0.008788594976067543,
-0.02192017063498497,
-0.025947261601686478,
0.07180564105510712,
-0.04538172110915184,
-0.03892558813095093,
0.06924939155578613,
-0.03321047127246857,
0.09376552700996399,
0.006... | 0.014665 |
{/\* This page is auto-generated from the skill's SKILL.md by website/scripts/generate-skill-docs.py. Edit the source SKILL.md, not this page. \*/} # Outlines Outlines: structured JSON/regex/Pydantic LLM generation. ## Skill metadata | | | |---|---| | Source | Optional β install with `hermes skills install official/mlo... | https://github.com/NousResearch/hermes-agent/blob/main/website/docs/user-guide/skills/optional/mlops/mlops-inference-outlines.md | main | hermes-agent | [
-0.03481438755989075,
0.012750878930091858,
-0.005614744033664465,
-0.029999712482094765,
-0.048711828887462616,
-0.038570571690797806,
-0.02503502182662487,
0.0052977497689425945,
-0.07298391312360764,
0.042591389268636703,
0.0023970562033355236,
-0.011388959363102913,
0.05860701575875282,
... | 0.101428 |
= outlines.generate.json(model, Product) product = generator("Extract: iPhone 15, $999, available") # Guaranteed valid Product instance print(type(product)) # ``` #### Regex Generator ```python # Generate text matching regex generator = outlines.generate.regex( model, r"[0-9]{3}-[0-9]{3}-[0-9]{4}" # Phone number patter... | https://github.com/NousResearch/hermes-agent/blob/main/website/docs/user-guide/skills/optional/mlops/mlops-inference-outlines.md | main | hermes-agent | [
-0.10997617989778519,
0.07103638350963593,
0.009253120049834251,
-0.02107386477291584,
-0.040435533970594406,
0.040301185101270676,
0.029336024075746536,
0.0593363493680954,
-0.008963152766227722,
-0.0952134057879448,
0.05203762277960777,
-0.11016511917114258,
0.06285186111927032,
0.019966... | 0.020031 |
cooking """ profile = generator(prompt) print(profile.full\_name) print(profile.interests) # ["hiking", "photography", "cooking"] ``` ### Pattern 4: Multi-Entity Extraction ```python class Entity(BaseModel): name: str type: Literal["PERSON", "ORGANIZATION", "LOCATION"] class DocumentEntities(BaseModel): entities: list[... | https://github.com/NousResearch/hermes-agent/blob/main/website/docs/user-guide/skills/optional/mlops/mlops-inference-outlines.md | main | hermes-agent | [
0.0036876853555440903,
0.07152943313121796,
0.0278644859790802,
0.05607977509498596,
0.03703361749649048,
-0.02243207022547722,
0.028039757162332535,
0.04030347988009453,
-0.008952834643423557,
-0.0355796255171299,
0.042154815047979355,
-0.09430801868438721,
0.013739374466240406,
-0.017270... | 0.019317 |
Comparison to Alternatives | Feature | Outlines | Instructor | Guidance | LMQL | |---------|----------|------------|----------|------| | Pydantic Support | β
Native | β
Native | β No | β No | | JSON Schema | β
Yes | β
Yes | β οΈ Limited | β
Yes | | Regex Constraints | β
Yes | β No | β
Yes | β
Yes | | Local Models | β
Ful... | https://github.com/NousResearch/hermes-agent/blob/main/website/docs/user-guide/skills/optional/mlops/mlops-inference-outlines.md | main | hermes-agent | [
0.01479707658290863,
-0.04410073533654213,
0.0006307339644990861,
-0.0018292948370799422,
0.012605395168066025,
-0.013474702835083008,
-0.14276981353759766,
0.10757430642843246,
-0.02755020000040531,
-0.007338258903473616,
0.04649001732468605,
-0.050398774445056915,
0.06497878581285477,
0.... | 0.058792 |
{/\* This page is auto-generated from the skill's SKILL.md by website/scripts/generate-skill-docs.py. Edit the source SKILL.md, not this page. \*/} # Stable Diffusion Image Generation State-of-the-art text-to-image generation with Stable Diffusion models via HuggingFace Diffusers. Use when generating images from text p... | https://github.com/NousResearch/hermes-agent/blob/main/website/docs/user-guide/skills/optional/mlops/mlops-stable-diffusion.md | main | hermes-agent | [
-0.053434163331985474,
0.0009412741055712104,
0.0002063327410724014,
0.01947781629860401,
0.030076637864112854,
-0.09082083404064178,
-0.06179291382431984,
-0.02791282720863819,
-0.07648813724517822,
0.021263837814331055,
0.047346875071525574,
-0.011716034263372421,
0.04809985309839249,
-0... | 0.117471 |
| Default choice | | `EulerAncestralDiscreteScheduler` | 20-50 | Good | More variation | | `DPMSolverMultistepScheduler` | 15-25 | Excellent | Fast, high quality | | `DDIMScheduler` | 50-100 | Good | Deterministic | | `LCMScheduler` | 4-8 | Good | Very fast | | `UniPCMultistepScheduler` | 15-25 | Excellent | Fast conve... | https://github.com/NousResearch/hermes-agent/blob/main/website/docs/user-guide/skills/optional/mlops/mlops-stable-diffusion.md | main | hermes-agent | [
-0.037795037031173706,
-0.01470211148262024,
0.02279762178659439,
0.025674765929579735,
-0.025726016610860825,
-0.07722881436347961,
-0.06614095717668533,
0.07285722345113754,
-0.057308729737997055,
-0.02716097980737686,
-0.011010723188519478,
-0.04975411668419838,
0.00002996226066898089,
... | -0.012184 |
memory by computing attention in chunks pipe.enable\_attention\_slicing() # Or specific chunk size pipe.enable\_attention\_slicing("max") ``` ### xFormers memory-efficient attention ```python # Requires xformers package pipe.enable\_xformers\_memory\_efficient\_attention() ``` ### VAE slicing for large images ```python... | https://github.com/NousResearch/hermes-agent/blob/main/website/docs/user-guide/skills/optional/mlops/mlops-stable-diffusion.md | main | hermes-agent | [
0.029250506311655045,
-0.01695677451789379,
-0.04249202832579613,
0.022300785407423973,
-0.01790049485862255,
-0.09250876307487488,
-0.012946917675435543,
0.10157202929258347,
-0.06496042013168335,
-0.06568653136491776,
-0.011811906471848488,
-0.015515376813709736,
-0.08398031443357468,
0.... | 0.091012 |
{/\* This page is auto-generated from the skill's SKILL.md by website/scripts/generate-skill-docs.py. Edit the source SKILL.md, not this page. \*/} # Huggingface Tokenizers Fast tokenizers optimized for research and production. Rust-based implementation tokenizes 1GB in <20 seconds. Supports BPE, WordPiece, and Unigram... | https://github.com/NousResearch/hermes-agent/blob/main/website/docs/user-guide/skills/optional/mlops/mlops-huggingface-tokenizers.md | main | hermes-agent | [
-0.10744891315698624,
0.022139281034469604,
-0.008781811222434044,
0.03894292190670967,
-0.05078969895839691,
-0.06430688500404358,
-0.014890900813043118,
0.0221693217754364,
-0.1000153124332428,
-0.017741497606039047,
-0.008849069476127625,
-0.012572216801345348,
0.06777448952198029,
-0.0... | 0.20641 |
special\_tokens=["<|endoftext|>"], min\_frequency=2 ) tokenizer.train(files=["data.txt"], trainer=trainer) ``` \*\*Advantages\*\*: - Handles OOV words well (breaks into subwords) - Flexible vocabulary size - Good for morphologically rich languages \*\*Trade-offs\*\*: - Tokenization depends on merge order - May split co... | https://github.com/NousResearch/hermes-agent/blob/main/website/docs/user-guide/skills/optional/mlops/mlops-huggingface-tokenizers.md | main | hermes-agent | [
-0.03898387402296066,
-0.01933452859520912,
0.03798595443367958,
0.032338980585336685,
-0.053439781069755554,
0.009320042096078396,
0.07140184938907623,
0.05919310823082924,
-0.041299156844615936,
-0.012165199965238571,
0.05573815852403641,
0.021716199815273285,
0.0873095691204071,
0.05481... | 0.111037 |
cases\*\*: - Named entity recognition (map predictions back to text) - Question answering (extract answer spans) - Token classification (align labels to original positions) ## Integration with transformers ### Load with AutoTokenizer ```python from transformers import AutoTokenizer # AutoTokenizer automatically uses fa... | https://github.com/NousResearch/hermes-agent/blob/main/website/docs/user-guide/skills/optional/mlops/mlops-huggingface-tokenizers.md | main | hermes-agent | [
-0.10297922044992447,
0.015537336468696594,
0.054873570799827576,
0.03654051944613457,
-0.007436312269419432,
0.04701247438788414,
0.05444195866584778,
0.05213245749473572,
-0.03823329880833626,
-0.04812825471162796,
0.02780638262629509,
-0.057559218257665634,
0.006119768135249615,
0.00748... | 0.119494 |
Nakajima, 2012) | https://github.com/NousResearch/hermes-agent/blob/main/website/docs/user-guide/skills/optional/mlops/mlops-huggingface-tokenizers.md | main | hermes-agent | [
-0.02459651231765747,
0.09333030134439468,
-0.014233329333364964,
0.055307786911726,
-0.06643997877836227,
0.008640624582767487,
-0.01037907600402832,
-0.016411207616329193,
-0.03430270031094551,
0.026464134454727173,
0.0666833221912384,
-0.05008084326982498,
0.016814805567264557,
0.016264... | 0.221882 |
{/\* This page is auto-generated from the skill's SKILL.md by website/scripts/generate-skill-docs.py. Edit the source SKILL.md, not this page. \*/} # Clip OpenAI's model connecting vision and language. Enables zero-shot image classification, image-text matching, and cross-modal retrieval. Trained on 400M image-text pai... | https://github.com/NousResearch/hermes-agent/blob/main/website/docs/user-guide/skills/optional/mlops/mlops-clip.md | main | hermes-agent | [
-0.04652911052107811,
-0.02624000795185566,
-0.02875939942896366,
-0.016293399035930634,
0.038568418473005295,
-0.08959227800369263,
-0.0068076010793447495,
-0.03185412660241127,
-0.03414376080036163,
-0.039736807346343994,
0.051356278359889984,
-0.0331059992313385,
0.037444379180669785,
0... | 0.173551 |
Content moderation ```python # Define categories categories = [ "safe for work", "not safe for work", "violent content", "graphic content" ] text = clip.tokenize(categories).to(device) # Check image with torch.no\_grad(): logits\_per\_image, \_ = model(image, text) probs = logits\_per\_image.softmax(dim=-1) # Get class... | https://github.com/NousResearch/hermes-agent/blob/main/website/docs/user-guide/skills/optional/mlops/mlops-clip.md | main | hermes-agent | [
0.04339773580431938,
-0.029278073459863663,
-0.02415088377892971,
0.03454238548874855,
0.0843629315495491,
-0.0668434277176857,
0.05256318673491478,
0.015330695547163486,
-0.03734643757343292,
-0.058819688856601715,
0.06243259087204933,
-0.019199686124920845,
0.021833062171936035,
0.034233... | 0.055737 |
{/\* This page is auto-generated from the skill's SKILL.md by website/scripts/generate-skill-docs.py. Edit the source SKILL.md, not this page. \*/} # Unsloth Unsloth: 2-5x faster LoRA/QLoRA fine-tuning, less VRAM. ## Skill metadata | | | |---|---| | Source | Optional β install with `hermes skills install official/mlops... | https://github.com/NousResearch/hermes-agent/blob/main/website/docs/user-guide/skills/optional/mlops/mlops-training-unsloth.md | main | hermes-agent | [
-0.11925617605447769,
-0.005575714632868767,
-0.05186532065272331,
0.026203544810414314,
-0.01930754818022251,
-0.055515632033348083,
-0.023036284372210503,
-0.05072024092078209,
-0.09216054528951645,
-0.03916526213288307,
0.00200386019423604,
-0.03426778316497803,
0.03809768706560135,
-0.... | 0.143882 |
{/\* This page is auto-generated from the skill's SKILL.md by website/scripts/generate-skill-docs.py. Edit the source SKILL.md, not this page. \*/} # Axolotl Axolotl: YAML LLM fine-tuning (LoRA, DPO, GRPO). ## Skill metadata | | | |---|---| | Source | Optional β install with `hermes skills install official/mlops/axolot... | https://github.com/NousResearch/hermes-agent/blob/main/website/docs/user-guide/skills/optional/mlops/mlops-training-axolotl.md | main | hermes-agent | [
-0.08475958555936813,
-0.024427607655525208,
-0.0014949776232242584,
-0.02478201873600483,
-0.04413868486881256,
-0.1028389036655426,
0.04021262377500534,
-0.00825449451804161,
-0.07216894626617432,
-0.022296693176031113,
-0.017472052946686745,
-0.02043137326836586,
0.038216061890125275,
-... | 0.151017 |
Beginners Start with the getting\_started or tutorials reference files for foundational concepts. ### For Specific Features Use the appropriate category reference file (api, guides, etc.) for detailed information. ### For Code Examples The quick reference section above contains common patterns extracted from the offici... | https://github.com/NousResearch/hermes-agent/blob/main/website/docs/user-guide/skills/optional/mlops/mlops-training-axolotl.md | main | hermes-agent | [
-0.0655980259180069,
-0.029777470976114273,
-0.028453733772039413,
0.01871298998594284,
0.06524455547332764,
-0.01107591949403286,
0.03989974781870842,
0.0771375223994255,
-0.09908311069011688,
0.0054985955357551575,
-0.0026558786630630493,
-0.008667828515172005,
0.010624478571116924,
-0.0... | 0.133159 |
{/\* This page is auto-generated from the skill's SKILL.md by website/scripts/generate-skill-docs.py. Edit the source SKILL.md, not this page. \*/} # Chroma Open-source embedding database for AI applications. Store embeddings and metadata, perform vector and full-text search, filter by metadata. Simple 4-function API. ... | https://github.com/NousResearch/hermes-agent/blob/main/website/docs/user-guide/skills/optional/mlops/mlops-chroma.md | main | hermes-agent | [
-0.08622671663761139,
0.018151285126805305,
-0.04234270006418228,
0.01505658496171236,
-0.005482879932969809,
-0.00938956718891859,
-0.06676522642374039,
0.0330914743244648,
-0.08618400990962982,
0.014758773148059845,
-0.01706342212855816,
-0.050546497106552124,
0.08347396552562714,
-0.041... | 0.1691 |
Update document content collection.update( ids=["id1"], documents=["Updated content"], metadatas=[{"source": "updated"}] ) ``` ### 6. Delete documents ```python # Delete by IDs collection.delete(ids=["id1", "id2"]) # Delete with filter collection.delete( where={"source": "outdated"} ) ``` ## Persistent storage ```pytho... | https://github.com/NousResearch/hermes-agent/blob/main/website/docs/user-guide/skills/optional/mlops/mlops-chroma.md | main | hermes-agent | [
-0.06707032769918442,
0.03423989936709404,
-0.003731353208422661,
0.05317939445376396,
0.00009350929030915722,
0.006177220959216356,
-0.008853377774357796,
0.023254279047250748,
0.05260326713323593,
-0.0036396048963069916,
0.005124363582581282,
0.005516488570719957,
0.10675570368766785,
-0... | -0.066544 |
{/\* This page is auto-generated from the skill's SKILL.md by website/scripts/generate-skill-docs.py. Edit the source SKILL.md, not this page. \*/} # Distributed Llm Pretraining Torchtitan Provides PyTorch-native distributed LLM pretraining using torchtitan with 4D parallelism (FSDP2, TP, PP, CP). Use when pretraining ... | https://github.com/NousResearch/hermes-agent/blob/main/website/docs/user-guide/skills/optional/mlops/mlops-torchtitan.md | main | hermes-agent | [
-0.07297921180725098,
-0.05017688870429993,
-0.032061439007520676,
-0.02300260215997696,
-0.03422456979751587,
-0.04842472821474075,
-0.0918322205543518,
0.021381612867116928,
-0.09673487395048141,
-0.018709851428866386,
-0.0013928775442764163,
0.007393032778054476,
-0.020447658374905586,
... | 0.068824 |
SLURM script\*\* ```bash #!/bin/bash #SBATCH --job-name=llama70b #SBATCH --nodes=32 #SBATCH --ntasks-per-node=8 #SBATCH --gpus-per-node=8 srun torchrun \ --nnodes=32 \ --nproc\_per\_node=8 \ --rdzv\_backend=c10d \ --rdzv\_endpoint=$MASTER\_ADDR:$MASTER\_PORT \ -m torchtitan.train \ --job.config\_file ./llama3\_70b.toml... | https://github.com/NousResearch/hermes-agent/blob/main/website/docs/user-guide/skills/optional/mlops/mlops-torchtitan.md | main | hermes-agent | [
-0.059417758136987686,
-0.054067209362983704,
-0.06794989109039307,
-0.0021537269931286573,
0.03535072132945061,
0.006046243477612734,
-0.03398814797401428,
-0.004528885707259178,
-0.11291693896055222,
-0.06705766171216965,
0.008221880532801151,
-0.0030426003504544497,
-0.02117123454809189,
... | 0.085689 |
Experimental | | Flux | Diffusion | Experimental | ## Performance benchmarks (H100) | Model | GPUs | Parallelism | TPS/GPU | Techniques | |-------|------|-------------|---------|------------| | Llama 8B | 8 | FSDP | 5,762 | Baseline | | Llama 8B | 8 | FSDP+compile+FP8 | 8,532 | +48% | | Llama 70B | 256 | FSDP+TP+AsyncT... | https://github.com/NousResearch/hermes-agent/blob/main/website/docs/user-guide/skills/optional/mlops/mlops-torchtitan.md | main | hermes-agent | [
-0.06861327588558197,
-0.003704000962898135,
-0.11148007959127426,
-0.006988188251852989,
0.0046967556700110435,
-0.0994495153427124,
-0.09679573029279709,
0.09082512557506561,
-0.1131771057844162,
0.03340812027454376,
0.021191328763961792,
-0.008824645541608334,
0.008580517023801804,
-0.0... | 0.202859 |
{/\* This page is auto-generated from the skill's SKILL.md by website/scripts/generate-skill-docs.py. Edit the source SKILL.md, not this page. \*/} # Slime Rl Training Provides guidance for LLM post-training with RL using slime, a Megatron+SGLang framework. Use when training GLM models, implementing custom data generat... | https://github.com/NousResearch/hermes-agent/blob/main/website/docs/user-guide/skills/optional/mlops/mlops-slime.md | main | hermes-agent | [
-0.09323135018348694,
-0.05616714060306549,
0.012779571115970612,
0.0012602424249053001,
-0.029857367277145386,
-0.026007749140262604,
-0.04452195391058922,
0.05065080150961876,
-0.10718841850757599,
-0.03756905719637871,
0.019918600097298622,
-0.04025416076183319,
0.05101033300161362,
-0.... | 0.062172 |
4"} ``` Or with chat format: ```python { "prompt": [ {"role": "system", "content": "You are a math tutor."}, {"role": "user", "content": "What is 15 + 27?"} ], "label": "42" } ``` ### Step 2: Configure Model Choose a pre-configured model script: ```bash # List available models ls scripts/models/ # glm4-9B.sh, qwen3-4B.... | https://github.com/NousResearch/hermes-agent/blob/main/website/docs/user-guide/skills/optional/mlops/mlops-slime.md | main | hermes-agent | [
-0.028286950662732124,
-0.019457412883639336,
-0.03524878993630409,
-0.06434906274080276,
-0.07578299939632416,
0.0014963941648602486,
-0.06369799375534058,
0.060793567448854446,
-0.03288650885224342,
0.006289452780038118,
0.006723857019096613,
-0.08340474963188171,
0.04579458013176918,
0.... | 0.006835 |
buffer\_filter(self, args, buffer, num\_samples): """Custom selection logic (prioritized, stratified, etc.).""" return select\_best(buffer, num\_samples) ``` --- ## Common Issues and Solutions ### Issue: SGLang Engine Crash \*\*Symptoms\*\*: Inference engine dies mid-training \*\*Solutions\*\*: ```bash # Enable fault t... | https://github.com/NousResearch/hermes-agent/blob/main/website/docs/user-guide/skills/optional/mlops/mlops-slime.md | main | hermes-agent | [
-0.027771739289164543,
-0.007183106616139412,
-0.015905002132058144,
0.021341197192668915,
0.004075319040566683,
-0.004474516026675701,
0.06834793090820312,
0.0630606934428215,
-0.11705447733402252,
-0.035381805151700974,
0.0009624089580029249,
-0.05289985612034798,
-0.00722916703671217,
-... | 0.129603 |
{/\* This page is auto-generated from the skill's SKILL.md by website/scripts/generate-skill-docs.py. Edit the source SKILL.md, not this page. \*/} # Mcporter Use the mcporter CLI to list, configure, auth, and call MCP servers/tools directly (HTTP or stdio), including ad-hoc servers, config edits, and CLI/type generati... | https://github.com/NousResearch/hermes-agent/blob/main/website/docs/user-guide/skills/optional/mcp/mcp-mcporter.md | main | hermes-agent | [
-0.07883447408676147,
-0.00912114605307579,
-0.030635658651590347,
0.00550189008936286,
-0.008696380071341991,
-0.08252131938934326,
0.03378590568900108,
-0.029636923223733902,
-0.07882516086101532,
0.0156675074249506,
0.04653977230191231,
0.0029952465556561947,
0.07376085966825485,
0.0005... | 0.102999 |
{/\* This page is auto-generated from the skill's SKILL.md by website/scripts/generate-skill-docs.py. Edit the source SKILL.md, not this page. \*/} # Fastmcp Build, test, inspect, install, and deploy MCP servers with FastMCP in Python. Use when creating a new MCP server, wrapping an API or database as MCP tools, exposi... | https://github.com/NousResearch/hermes-agent/blob/main/website/docs/user-guide/skills/optional/mcp/mcp-fastmcp.md | main | hermes-agent | [
-0.0928344801068306,
-0.015369903296232224,
-0.029562579467892647,
-0.003922830335795879,
-0.046699367463588715,
-0.09261339157819748,
-0.019373023882508278,
-0.03548441454768181,
-0.0781092420220375,
0.015588064678013325,
0.0665423721075058,
0.0012295308988541365,
0.10981893539428711,
-0.... | 0.060507 |
descriptions - Keep parameters explicit and typed - Return structured JSON-safe data where possible - Validate unsafe inputs early - Prefer read-only behavior by default for first versions Good tool examples: - `get\_customer` - `search\_tickets` - `describe\_table` - `summarize\_text\_file` Weak tool examples: - `run`... | https://github.com/NousResearch/hermes-agent/blob/main/website/docs/user-guide/skills/optional/mcp/mcp-fastmcp.md | main | hermes-agent | [
-0.030131803825497627,
0.04344608262181282,
-0.037054482847452164,
-0.0139393862336874,
-0.003322842065244913,
-0.0090249702334404,
-0.0007394092390313745,
0.08611399680376053,
-0.01890212669968605,
0.00008720301411813125,
0.009839185513556004,
-0.005110979080200195,
0.08584006130695343,
-... | 0.086063 |
FastMCP server, verify all of the following: - server imports cleanly - `fastmcp inspect ` succeeds - `fastmcp list --json` succeeds - every new tool has at least one real `fastmcp call` - environment variables are documented - the tool surface is small enough to understand without guesswork ## Troubleshooting ### Fast... | https://github.com/NousResearch/hermes-agent/blob/main/website/docs/user-guide/skills/optional/mcp/mcp-fastmcp.md | main | hermes-agent | [
-0.07740883529186249,
0.014423185028135777,
0.042175356298685074,
-0.04869518056511879,
0.020499669015407562,
-0.11063093692064285,
-0.06741467863321304,
-0.0398673489689827,
-0.009730118326842785,
-0.016404107213020325,
0.07802064716815948,
-0.04029829800128937,
0.036112044006586075,
0.03... | -0.053143 |
{/\* This page is auto-generated from the skill's SKILL.md by website/scripts/generate-skill-docs.py. Edit the source SKILL.md, not this page. \*/} # Hyperliquid Hyperliquid market data, account history, trade review. ## Skill metadata | | | |---|---| | Source | Optional β install with `hermes skills install official/b... | https://github.com/NousResearch/hermes-agent/blob/main/website/docs/user-guide/skills/optional/blockchain/blockchain-hyperliquid.md | main | hermes-agent | [
-0.06268490105867386,
0.04106029123067856,
-0.02596980705857277,
0.00813696626573801,
-0.03666042163968086,
-0.0700535997748375,
0.06394830346107483,
-0.06905676424503326,
-0.02613190934062004,
-0.01995643600821495,
0.04124363139271736,
-0.004400003235787153,
0.04634552821516991,
-0.027738... | 0.169518 |
book depth, near-term liquidity, or potential market impact of a large order. ### 4. Review an Account ```bash python3 ~/.hermes/skills/blockchain/hyperliquid/scripts/hyperliquid\_client.py \ state 0xabc... python3 ~/.hermes/skills/blockchain/hyperliquid/scripts/hyperliquid\_client.py \ spot-balances ``` `state` return... | https://github.com/NousResearch/hermes-agent/blob/main/website/docs/user-guide/skills/optional/blockchain/blockchain-hyperliquid.md | main | hermes-agent | [
-0.032206207513809204,
0.039216186851263046,
-0.0300757959485054,
0.0186162106692791,
-0.04207366332411766,
-0.048381514847278595,
0.027228476479649544,
0.002171759959310293,
0.004019768442958593,
0.007442186586558819,
0.0030662366189062595,
0.019442258402705193,
0.007199302315711975,
-0.0... | 0.07028 |
{/\* This page is auto-generated from the skill's SKILL.md by website/scripts/generate-skill-docs.py. Edit the source SKILL.md, not this page. \*/} # Evm Read-only EVM client: wallets, tokens, gas across 8 chains. ## Skill metadata | | | |---|---| | Source | Optional β install with `hermes skills install official/block... | https://github.com/NousResearch/hermes-agent/blob/main/website/docs/user-guide/skills/optional/blockchain/blockchain-evm.md | main | hermes-agent | [
-0.06562696397304535,
0.04924175143241882,
-0.06226399913430214,
-0.01745554432272911,
0.007952420972287655,
-0.07894986867904663,
0.016034668311476707,
-0.032924119383096695,
-0.06094278022646904,
0.00046857245615683496,
0.06667450815439224,
0.0030016687233000994,
0.08832835406064987,
-0.... | 0.084551 |
Gas python3 $SCRIPT gas # Gas prices + cost estimates python3 $SCRIPT gas --chain optimism # Security python3 $SCRIPT allowance 0xd8dA...96045 # Dangerous ERC-20 approvals python3 $SCRIPT contract 0xdAC1...1ec7 # Contract inspection (proxy? standards?) # ENS python3 $SCRIPT ens vitalik.eth # Name -> address + profile p... | https://github.com/NousResearch/hermes-agent/blob/main/website/docs/user-guide/skills/optional/blockchain/blockchain-evm.md | main | hermes-agent | [
-0.019926078617572784,
0.07780973613262177,
-0.058074988424777985,
0.009540469385683537,
0.016429411247372627,
-0.1206355020403862,
0.025386299937963486,
0.027591295540332794,
-0.0742422342300415,
0.04210706800222397,
-0.01205383986234665,
-0.10477469116449356,
0.04220055416226387,
-0.0065... | 0.067253 |
public endpoint (4byte.directory) with no fallback. Selectors not in their database show up as `unknown`. - \*\*L2 gas estimates are L2-execution only.\*\* On rollups like Base, Arbitrum, Optimism, and zkSync, the actual transaction cost also includes an L1 data-posting fee that depends on calldata size and current L1 ... | https://github.com/NousResearch/hermes-agent/blob/main/website/docs/user-guide/skills/optional/blockchain/blockchain-evm.md | main | hermes-agent | [
-0.021725859493017197,
0.029568227007985115,
-0.08767999708652496,
0.049278222024440765,
0.05187072977423668,
-0.10561437159776688,
-0.005906532518565655,
0.09657129645347595,
-0.011501738801598549,
-0.0017274118727073073,
0.033273253589868546,
-0.026056300848722458,
0.028601525351405144,
... | 0.020657 |
{/\* This page is auto-generated from the skill's SKILL.md by website/scripts/generate-skill-docs.py. Edit the source SKILL.md, not this page. \*/} # Solana Query Solana blockchain data with USD pricing β wallet balances, token portfolios with values, transaction details, NFTs, whale detection, and live network stats. ... | https://github.com/NousResearch/hermes-agent/blob/main/website/docs/user-guide/skills/optional/blockchain/blockchain-solana.md | main | hermes-agent | [
-0.11468879133462906,
0.020043564960360527,
-0.1301514357328415,
0.019211914390325546,
-0.03014778345823288,
-0.11098501831293106,
-0.021375343203544617,
-0.0166146382689476,
-0.05641268566250801,
0.02770346775650978,
0.007281861733645201,
-0.02778349071741104,
0.014810673892498016,
-0.028... | 0.16607 |
token DezXAZ8z7PnrnRJjz3wXBoRgixCa6xjnB7YaB1pPB263 ``` Output: name, symbol, decimals, supply, price, market cap, top 5 holders with percentages. ### 4. Recent Activity List recent transactions for an address (default: last 10, max: 25). ```bash python3 ~/.hermes/skills/blockchain/solana/scripts/solana\_client.py \ act... | https://github.com/NousResearch/hermes-agent/blob/main/website/docs/user-guide/skills/optional/blockchain/blockchain-solana.md | main | hermes-agent | [
-0.023031610995531082,
0.03973481431603432,
-0.06896975636482239,
-0.027087610214948654,
0.0330401174724102,
-0.08088710159063339,
0.10145914554595947,
0.009687593206763268,
0.031390849500894547,
0.012458045035600662,
0.010238430462777615,
-0.0036391892936080694,
0.001335171633400023,
0.00... | 0.14292 |
{/\* This page is auto-generated from the skill's SKILL.md by website/scripts/generate-skill-docs.py. Edit the source SKILL.md, not this page. \*/} # Dcf Model Build institutional-quality DCF valuation models in Excel β revenue projections, FCF build, WACC, terminal value, Bear/Base/Bull scenarios, 5x5 sensitivity tabl... | https://github.com/NousResearch/hermes-agent/blob/main/website/docs/user-guide/skills/optional/finance/finance-dcf-model.md | main | hermes-agent | [
-0.015271101146936417,
0.0005849653389304876,
-0.1555648297071457,
-0.013325173407793045,
-0.0275914017111063,
-0.001104639028199017,
-0.08150625973939896,
0.026431947946548462,
-0.015343726612627506,
0.06202816218137741,
0.018855402246117592,
-0.07255598157644272,
0.0408363975584507,
-0.0... | 0.029752 |
is 3.0%). The center cell's output must therefore equal the model's actual implied share price β this is the sanity check that the table is built correctly. - \*\*Highlight the center cell\*\* with the medium-blue fill (`#BDD7EE`) + bold font so it's immediately visible which cell is the base case. - Populate ALL cells... | https://github.com/NousResearch/hermes-agent/blob/main/website/docs/user-guide/skills/optional/finance/finance-dcf-model.md | main | hermes-agent | [
-0.02407691814005375,
-0.023272644728422165,
-0.048630956560373306,
0.04387475177645683,
-0.0015964409103617072,
0.05415167286992073,
-0.06238820031285286,
0.06737722456455231,
-0.038700442761182785,
0.08390528708696365,
0.01797196827828884,
-0.12370262295007706,
0.07275339215993881,
-0.02... | 0.001598 |
(e.g., 16-20%) ``` ### Step 4: Operating Expense Modeling \*\*Fixed/Variable Cost Analysis:\*\* Operating expenses should model realistic operating leverage: - \*\*Sales & Marketing\*\*: Typically 15-40% of revenue depending on business model - \*\*Research & Development\*\*: Typically 10-30% for technology companies -... | https://github.com/NousResearch/hermes-agent/blob/main/website/docs/user-guide/skills/optional/finance/finance-dcf-model.md | main | hermes-agent | [
0.017186596989631653,
0.01841568946838379,
-0.05907077342271805,
-0.0006308044539764524,
-0.03705337271094322,
-0.014918596483767033,
-0.027109140530228615,
0.06817701458930969,
-0.033389586955308914,
0.022742362692952156,
-0.02080581523478031,
-0.03468023240566254,
0.009766308590769768,
-... | 0.058785 |
Standard for most analyses - \*\*7-10 years\*\*: High growth companies with longer runway - \*\*3 years\*\*: Mature, stable businesses ### Step 8: Terminal Value Calculation \*\*Perpetuity Growth Method (Preferred):\*\* ``` Terminal FCF = Final Year FCF Γ (1 + Terminal Growth Rate) Terminal Value = Terminal FCF / (WACC... | https://github.com/NousResearch/hermes-agent/blob/main/website/docs/user-guide/skills/optional/finance/finance-dcf-model.md | main | hermes-agent | [
0.0930033028125763,
0.0023554961662739515,
-0.06980782747268677,
-0.05686953291296959,
-0.0015513234538957477,
0.034507907927036285,
-0.06222502142190933,
0.10722078382968903,
-0.02379695512354374,
0.08651148527860641,
0.07603593170642853,
-0.07663828134536743,
0.00905323214828968,
-0.0374... | 0.06034 |
CASE ASSUMPTIONS (section header, merge cells across) Assumption,FY1,FY2,FY3,FY4,FY5 Revenue Growth (%),20%,18%,15%,13%,11% EBIT Margin (%),50%,51%,52%,53%,54% ``` \*\*Each scenario block MUST have a column header row\*\* showing the projection years (FY2025E, FY2026E, etc.) immediately below the section title. Without... | https://github.com/NousResearch/hermes-agent/blob/main/website/docs/user-guide/skills/optional/finance/finance-dcf-model.md | main | hermes-agent | [
0.001158117433078587,
0.06418782472610474,
-0.004440422169864178,
-0.04071672633290291,
0.023006621748209,
0.07982504367828369,
-0.06039819121360779,
0.013282633386552334,
-0.06191384792327881,
0.06159893795847893,
0.01404499914497137,
-0.09181202948093414,
0.03203053027391434,
-0.01054369... | 0.015 |
case selector. This consolidation column is what your projection formulas reference. ### Correct Row Planning Process \*\*1. Write ALL headers and labels FIRST:\*\* ```csv Row,Content 1,[Company Name] DCF Model 2,Ticker | Date | Year End 4,Case Selector 7,KEY ASSUMPTIONS 26,Assumption headers 27-31,Growth assumptions .... | https://github.com/NousResearch/hermes-agent/blob/main/website/docs/user-guide/skills/optional/finance/finance-dcf-model.md | main | hermes-agent | [
-0.002370405476540327,
-0.002071186201646924,
-0.041110727936029434,
-0.007415829226374626,
0.04782174900174141,
0.02689170651137829,
-0.10633665323257446,
0.023128865286707878,
-0.027872247621417046,
0.09314580261707306,
0.022178050130605698,
-0.13373839855194092,
0.06282249838113785,
-0.... | 0.032725 |
approximation B97: =B88\*(1+(0.096-0.116)) // Assumes linear relationship // WRONG - Division shortcut B105: =B88/(1+(E48-0.07)) // Doesn't recalculate full DCF ``` \*\*Don't leave placeholder text:\*\* ``` // WRONG - Placeholder note "Note: Use Excel Data Table feature (Data β What-If Analysis β Data Table) to populat... | https://github.com/NousResearch/hermes-agent/blob/main/website/docs/user-guide/skills/optional/finance/finance-dcf-model.md | main | hermes-agent | [
0.010349194519221783,
0.021706953644752502,
0.017790790647268295,
0.031067630276083946,
-0.022392084822058678,
-0.03084181807935238,
-0.054085079580545425,
0.1091005876660347,
-0.043648798018693924,
0.05086907744407654,
0.06586823612451553,
-0.14222563803195953,
0.04824327677488327,
-0.005... | -0.025162 |
vs black \*\*Why it's wrong:\*\* - Can't distinguish inputs from formulas - Auditing becomes impossible - Violates xlsx skill requirements \*\*Instead:\*\* Blue text for ALL hardcoded inputs, black text for ALL formulas, green for sheet links ### WRONG: Operating Expenses Based on Gross Profit \*\*Don't do this:\*\* `S... | https://github.com/NousResearch/hermes-agent/blob/main/website/docs/user-guide/skills/optional/finance/finance-dcf-model.md | main | hermes-agent | [
0.010554462671279907,
0.011009672656655312,
-0.014716805890202522,
-0.003554258495569229,
0.01727115735411644,
0.03778404742479324,
-0.014271696098148823,
0.02385592646896839,
0.05305405333638191,
0.0433146171271801,
0.053316351026296616,
-0.10309262573719025,
0.08185357600450516,
-0.01206... | 0.034567 |
Sensitivity tables go at the BOTTOM of the DCF sheet (not on a separate sheet). This keeps all valuation outputs together. ### Formula Recalculation (MANDATORY) After creating or modifying the Excel model, \*\*recalculate all formulas\*\* using the `recalc.py` script from the `excel-author` skill: ```bash python recalc... | https://github.com/NousResearch/hermes-agent/blob/main/website/docs/user-guide/skills/optional/finance/finance-dcf-model.md | main | hermes-agent | [
0.015377587638795376,
-0.021059388294816017,
-0.07932563871145248,
0.009239805862307549,
-0.003395745065063238,
-0.010822047479450703,
-0.09220089018344879,
0.1203819140791893,
0.033981554210186005,
-0.017861422151327133,
0.023180587217211723,
-0.09540247917175293,
0.03840997442603111,
-0.... | -0.048192 |
\*\*No borders:\*\* Individual cells within tables (keep clean, scannable) \*\*Borders are mandatory\*\* - models without professional borders are not client-ready. \*\*Number Formats\*\* (follows xlsx skill standards): - \*\*Years\*\*: Format as text strings (e.g., "2024" not "2,024") - \*\*Percentages\*\*: `0.0%` (on... | https://github.com/NousResearch/hermes-agent/blob/main/website/docs/user-guide/skills/optional/finance/finance-dcf-model.md | main | hermes-agent | [
0.032064519822597504,
0.08486229181289673,
-0.06499653309583664,
-0.019347412511706352,
-0.044802043586969376,
0.054457489401102066,
-0.025332944467663765,
0.002862317720428109,
0.035009194165468216,
0.05269902944564819,
0.04266825690865517,
-0.12604236602783203,
0.08937735855579376,
-0.00... | -0.026956 |
Risk-Free Rate (10Y Treasury),X.XX%,[Yellow input] Beta (5Y monthly),X.XX,[Yellow input] Equity Risk Premium,X.XX%,[Yellow input] Cost of Equity,X.XX%,[Calculated blue] ,, COST OF DEBT CALCULATION,, Credit Rating,AA-,[Yellow input] Pre-Tax Cost of Debt,X.XX%,[Yellow input] Tax Rate,XX.X%,[Link to DCF sheet] After-Tax C... | https://github.com/NousResearch/hermes-agent/blob/main/website/docs/user-guide/skills/optional/finance/finance-dcf-model.md | main | hermes-agent | [
-0.015399767085909843,
-0.01766885071992874,
-0.10424961894750595,
-0.05492238327860832,
0.019455591216683388,
0.04926437512040138,
0.021437542513012886,
0.07679923623800278,
-0.007337242364883423,
0.0751301720738411,
0.06280047446489334,
-0.05387174338102341,
-0.0026176092214882374,
-0.02... | 0.002977 |
pull from the appropriate scenario block. \*\*Recommended pattern (using INDEX):\*\* `=INDEX(B10:D10, 1, $B$6)` where `B10:D10` = Bear/Base/Bull values, `1` = row offset, `$B$6` = case selector cell (1, 2, or 3) \*\*Then reference the consolidation column\*\* in all projections: `Revenue Year 1: =D29\*(1+$E$10)` where ... | https://github.com/NousResearch/hermes-agent/blob/main/website/docs/user-guide/skills/optional/finance/finance-dcf-model.md | main | hermes-agent | [
-0.02474852092564106,
0.010581615380942822,
-0.022018957883119583,
0.024402091279625893,
0.011919815093278885,
0.06466945260763168,
-0.040304068475961685,
0.05812416598200798,
-0.005996187217533588,
0.07371441274881363,
0.03145342320203781,
-0.048353809863328934,
0.03956730291247368,
-0.04... | 0.001228 |
sheet) - Font colors: Blue inputs, black formulas, green sheet links - Cell comments on ALL hardcoded inputs - Professional borders around major sections 2. \*\*Recalculate formulas\*\*: Run `python recalc.py model.xlsx 30` 3. \*\*Check output\*\*: - If `status` is `"success"` β Continue to step 4 - If `status` is `"er... | https://github.com/NousResearch/hermes-agent/blob/main/website/docs/user-guide/skills/optional/finance/finance-dcf-model.md | main | hermes-agent | [
-0.05338525027036667,
-0.01325762178748846,
-0.048703018575906754,
-0.012131886556744576,
-0.010200316086411476,
0.010992903262376785,
-0.03996174409985542,
0.026525219902396202,
-0.014958816580474377,
0.0075891767628490925,
0.05881056562066078,
-0.03784336894750595,
0.041836097836494446,
... | 0.038051 |
{/\* This page is auto-generated from the skill's SKILL.md by website/scripts/generate-skill-docs.py. Edit the source SKILL.md, not this page. \*/} # Comps Analysis Build comparable company analysis in Excel β operating metrics, valuation multiples, statistical benchmarking vs peer sets. Pairs with excel-author. Use fo... | https://github.com/NousResearch/hermes-agent/blob/main/website/docs/user-guide/skills/optional/finance/finance-comps-analysis.md | main | hermes-agent | [
-0.0270399060100317,
0.05347876623272896,
-0.13535062968730927,
-0.03224993124604225,
-0.035030730068683624,
-0.03265472501516342,
-0.03184047341346741,
0.01937807723879814,
-0.02497543953359127,
0.03126811236143112,
0.0015616340097039938,
-0.023232758045196533,
0.02096271887421608,
-0.027... | 0.105466 |
principle:\*\* Use template principles (clear structure, statistical rigor, transparent formulas) but vary execution based on context. The goal is institutional-quality analysis, not institutional-looking templates. User-provided examples and explicit preferences always take precedence over defaults. ## Core Philosophy... | https://github.com/NousResearch/hermes-agent/blob/main/website/docs/user-guide/skills/optional/finance/finance-comps-analysis.md | main | hermes-agent | [
-0.02261173725128174,
0.08205004036426544,
-0.06077912449836731,
-0.014574546366930008,
-0.06901545822620392,
-0.001237324089743197,
-0.1045069620013237,
0.10871315747499466,
0.01678243838250637,
0.07180794328451157,
0.007759226486086845,
-0.08606993407011032,
0.017238618806004524,
0.03448... | 0.077149 |
for hardcoded inputs - \*\*Statistics rows\*\* (Maximum, 75th Percentile, etc.): - Light grey background (`#F2F2F2`) - Black text, left-aligned labels - \*\*That's the whole palette\*\*: dark blue + light blue + light grey + white. Nothing else unless the user's template says otherwise. \*\*Suggested Formatting Convent... | https://github.com/NousResearch/hermes-agent/blob/main/website/docs/user-guide/skills/optional/finance/finance-comps-analysis.md | main | hermes-agent | [
-0.009711310267448425,
0.06890031695365906,
-0.12150406837463379,
-0.010013296268880367,
-0.023467831313610077,
0.04963240027427673,
-0.0244658961892128,
-0.007837867364287376,
-0.0226762555539608,
0.03651395067572594,
-0.05781416967511177,
-0.08507712930440903,
0.04477756842970848,
-0.011... | 0.031732 |
per dollar of earnings 6. \*\*P/E Ratio\*\* - Price relative to net earnings ### Optional Valuation Metrics (Choose based on context) - \*\*FCF Yield\*\* - FCF/Market Cap (for cash-focused analysis) - \*\*PEG Ratio\*\* - P/E/Growth Rate (for growth companies) - \*\*Price/Book\*\* - Market value vs. book value (for asse... | https://github.com/NousResearch/hermes-agent/blob/main/website/docs/user-guide/skills/optional/finance/finance-comps-analysis.md | main | hermes-agent | [
0.0008033362682908773,
-0.02972719818353653,
-0.13363906741142273,
-0.042621910572052,
-0.013074934482574463,
0.02104841358959675,
-0.06765627861022949,
0.07662748545408249,
0.049296483397483826,
-0.010478550568223,
-0.005523218307644129,
-0.07915061712265015,
-0.019796352833509445,
-0.021... | 0.104728 |
banks) \*\*Retail/E-commerce:\*\* Must have: Revenue Growth, Gross Margin, Inventory Turnover Optional: Same-Store Sales, Customer Acquisition Cost Skip: Heavy R&D or CapEx metrics ### The "5-10 Rule" \*\*5 operating metrics\*\* - Revenue, Growth, 2-3 margins/efficiency metrics \*\*5 valuation metrics\*\* - Market Cap,... | https://github.com/NousResearch/hermes-agent/blob/main/website/docs/user-guide/skills/optional/finance/finance-comps-analysis.md | main | hermes-agent | [
0.03687440976500511,
-0.047202687710523605,
-0.02194708026945591,
-0.03344911336898804,
-0.03118486888706684,
-0.013032433576881886,
-0.004345157649368048,
-0.016460737213492393,
-0.0574369840323925,
-0.03357578068971634,
-0.03310491889715195,
-0.034878019243478775,
-0.019347617402672768,
... | 0.039965 |
Typical market valuation - \*\*25th percentile\*\* = "Discount" territory This helps answer: "Is our target company trading rich or cheap vs. peers?" ### Industry-Specific Modifications \*\*Software/SaaS:\*\* - Add: ARR, Net Dollar Retention, CAC Payback Period - Emphasize: Rule of 40, FCF margins, gross margins >70% \... | https://github.com/NousResearch/hermes-agent/blob/main/website/docs/user-guide/skills/optional/finance/finance-comps-analysis.md | main | hermes-agent | [
-0.008063470013439655,
0.009846865199506283,
-0.093597412109375,
-0.012709219940006733,
-0.010698452591896057,
0.07323042303323746,
-0.05698377639055252,
0.0674421489238739,
-0.016001097857952118,
0.06981105357408524,
-0.02519088424742222,
-0.0399487279355526,
0.03616160899400711,
-0.07147... | 0.099386 |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.